POINT-BASED IONIZING RADIATION DOSIMETRY USING RADIOCHROMIC MATERIALS AND A FIBREOPTIC READOUT SYSTEM

Transcription

1 POINT-BASED IONIZING RADIATION DOSIMETRY USING RADIOCHROMIC MATERIALS AND A FIBREOPTIC READOUT SYSTEM by Alexandra Rink A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Medical Biophysics University of Toronto Copyright by Alexandra Rink (2008)

2 Abstract Point-Based Ionizing Radiation Dosimetry Using Radiochromic Materials And Fibreoptic Readout System Doctor of Philosophy, 2008 Alexandra Rink Department of Medical Biophysics University of Toronto Real-time feedback of absorbed dose at a point within a patient can help with radiological quality assurance and innovation. Two radiochromic materials from GafChromic MD-55 and EBT films have been investigated for applicability in real-time in vivo dosimetry of ionizing radiation. Both films were able to produce a real-time measurement of optical density from a small volume, allowing positioning onto a tip of an optical fibre in the future. The increase in optical density was linear with absorbed dose for MD-55, and non-linear for EBT. The nonlinearity of EBT is associated with its increased sensitivity to ionizing radiation compared to MD-55, thus reaching optical saturation at a much lower dose. The radiochromic material in EBT film was also shown to polymerize and stabilize faster, decreasing dose rate dependence in real-time measurements in comparison to MD-55. The response of the two media was tested over 75 kv p 18 MV range of x-ray beams. The optical density measured for EBT was constant within 3% throughout the entire range, while MD-55 exhibited a nearly 40% decrease at low energies. Both materials were also shown to be temperature sensitive, with the change in optical density generally decreasing when the temperature increased from ~22 C to ~37 C. This was accompanied by a shift in the peak absorbance wavelength. It was illustrated that some of this decrease can be corrected for by tracking the peak position and then multiplying the optical density by a correction factor based on the predicted temperature. Overall, the radiochromic ii

3 material in GafChromic EBT film was found to be a better candidate for in vivo real-time dosimetry than the material in GafChromic MD-55. A novel mathematical model was proposed linking absorbance to physical parameters and processes of the radiochromic materials. The absorbance at every wavelength in the spectrum was represented as a sum of absorbances from multiple absorbers, where absorbance is characterized by its absorption coefficient, initiation constant, and polymerization constant. Preliminary fits of this model to experimental data assuming two absorbers suggested that there is a trade-off between EBT s greater sensitivity and its dose linearity characteristics. This was confirmed by experimental results. iii

4 Dedicated to my dear parents. iv

5 Acknowledgments The assistance and wisdom of many people went into this work. I would gratefully like to acknowledge the contributions, in whatever form they were, of the following: my co-supervisors, Dr. David Jaffray and Dr. Alex Vitkin, for all the paper edits, research guidance and advice committee members, Dr. Christine Allen and Dr. Mike Rauth for all the support Yuen Wong, Brian Taylor, Jason Ellis, and Matt Filletti for machining all the phantoms and doing the various small rush jobs Robert Rothwell and Robert Rusnov for all the assistance with electrical and optical work Dr. Robert Heaton, Hamideh Alasti, Duncan Galbraith, Dr. Mohammad Islam, and Dr. Jean-Pierre Bissonnette for their experience Bern Norrlinger for his experience and help with any and every accelerator that ever broke Tony Manfredi for all the assistance with the Elekta accelerators Dr. Robert Weersink and David Giewercer for assistance and wisdom with fibre optics Joanne Kniaz of Advanced Optical Microscopy Facility for the microscopy work Dr. David Lewis and Dr. Sangya Varma of International Specialty Products for their contributions to this work, experience and guidance Dr. Douglas Moseley for all the help with Matlab, the carpool rides, and outrageous conversations on the GO train Steve Ansell and Graham Wilson for all the computer support, psychotherapy lunches and Chinese noodles Jinzi Zheng and Jeremy Hoisak for all the coffee breaks which kept me sane my parents, Gala and Youri Rink, for never looking back or regretting any choices in life v

6 Table of Contents CHAPTER 1: INTRODUCTION... 1 I. Ionizing Radiation in Cancer Treatment... 2 II. Radiation Dose... 3 III. Radiation Dosimetry... 5 A. Basic Interactions... 5 B. Standards and Protocols for Dosimetry... 6 C. Estimation of Dose Delivered in Therapy... 7 IV. The Challenges of Dose Measurement in the Clinical Setting... 7 A. Clinical Applications and Ideal Dosimeter... 7 B. Current in vivo Dosimeters C. Optical Methods V. Outline of Thesis CHAPTER 2: REAL-TIME RESPONSE OF GAFCHROMIC MD-55 FILM TO IONIZING RADIATION I. Introduction Review of GafChromic MD General Experience Solid-state Polymerization of Diacetylenes II. Methods and Materials A. ΔOD of GafChromic MD-55 at Various Doses B. Sensitivity as a Function of Layer Thickness C. ΔOD of GafChromic MD-55 at Various Dose Rates D. ΔOD Dependency on Temperature E. Continuous Versus Pulsed Irradiation III. Results A. ΔOD of GafChromic MD-55 at Various Doses B. Sensitivity as a Function of Layer Thickness C. ΔOD of GafChromic MD-55 at Various Dose Rates D. ΔOD Dependency on Temperature E. Continuous Versus Pulsed Irradiation IV. Discussion vi

7 A. OD of GafChromic MD-55 at Various Doses B. Sensitivity as a Function of Layer Thickness C. ΔOD of GafChromic MD-55 at Various Dose Rates D. ΔOD Dependency on Temperature E. Applications V. Conclusion ACKNOWLEDGEMENTS CHAPTER 3: REAL-TIME RESPONSE OF GAFCHROMIC EBT I. Introduction II. Method and Materials A. ΔOD of EBT Film Versus Time B. Sensitivity and Stability Comparison Between EBT and MD-55 Films C. Dependence of Real-Time OD Measurements on Dose Rate for the EBT Film D. Structure of Active Crystals in MD-55 and EBT Films III. Results and Discussion A. OD of EBT Film Versus Time B. Sensitivity and Stability Comparison Between EBT and MD-55 Films C. Dependence of Real-Time ΔOD Measurements on Dose Rate for the EBT Film D. Structure of Active Crystals in MD-55 and EBT Films IV. Conclusion ACKNOWLEDGEMENTS CHAPTER 4: EFFECTS OF VARYING DOSE RATE ON REAL-TIME MEASUREMENTS OF OPTICAL DENSITY OF GAFCHROMIC EBT I. Introduction II. Methods and Materials III. Results and Discussion IV. Conclusion ACKNOWLEDGEMENTS CHAPTER 5: CHARACTERIZATION OF GAFCHROMIC EBT: TEMPERATURE AND HUMIDITY EFFECTS I. Introduction Chemical Background and General Experience II. Methods and Materials vii

8 A. Temperature Dependence B. Absorbance and Sensitivity Dependence on Water Content III. Results and Discussion A. Temperature Dependence B. Absorbance and Sensitivity Dependence on Water Content IV. Conclusion ACKNOWLEDGEMENTS CHAPTER 6: ENERGY DEPENDENCE OF GAFCHROMIC MD-55 AND EBT I. Introduction II. Methods and Materials Solid Water Phantom Ionizing Radiation Exposures Optical Measurements III. Results and Discussion IV. Conclusion ACKNOWLEDGEMENTS CHAPTER 7: MATHEMATICAL MODEL OF RADIOCHROMIC MEDIUM RESPONSE TO IONIZING RADIATION I. Introduction II. Methods and Materials III. Results and Discussion V. Conclusion ACKNOWLEDGEMENTS CHAPTER 8: SUMMARY AND FUTURE DIRECTIONS I. Summary II. Future Directions A. Optical Probe B. Organization of Monomers and Polymers C. Importance of Chemical Composition and Structure D. New Radiochromic Materials E. Polymerization Kinetics as a Function of Dose Per Pulse F. Model-Fitting Algorithm and Code viii

9 List of Tables Table 1. List of criteria for in vivo point-based real-time dosimeter Table 2. Evaluation criteria for in vivo point-based real-time dosimeter Table 3. Comparison of inferred dose and percent error using calibration plot and pre-exposure calibration as methods of calculation Table 4. Coefficients of equations of best fit characterizing ΔOD/D Gy as a function of dose rate Table 5. Average percent standard deviation for each dose, uncertainty, and the difference between the two for each dose delivered Table 6. Correction factors for the temperature correction scheme, calculated for doses of cgy shown for a selection of predicted temperature values Table 7. The x-ray and photon beams employed in the investigations Table 8. Comparison of response of EBT film, normalized to response at 6 MV, as measured approximately 24 hours after exposure to that measured immediately at the end of exposure Table 9. Model parameters using two absorbers for MD-55 and EBT and a fit with 1 second pulse averaging List of Figures Figure 1. Schematic of a typical relationship between tumour control probability (TCP) and normal tissue complication probability (NTCP) versus dose Figure 2. Structures of: (a) diacetylene monomers, upon exposure to ionizing radiation, polymerizes into (b) butatriene structure polymer; as the polymer chain grows, it rearranges via (c) an intermediate between butatriene structure and acetylene structure, into (d) acetylene structure polymer Figure 3. A model of optical density of GafChromic MD-55 versus time before, during and after exposure Figure 4. Schematic of experimental setup Figure 5. Emission spectrum of the LED as detected by the spectrophotometer Figure 6(a-c). Solid Water phantom (a) assembled, (b) disassembled, (c) schematic Figure 7. Schematic of cross-section of film holder in Solid Water phantom ix

10 Figure 8. Change in absorbance of GafChromic MD-55 film at various wavelengths plotted before exposure, immediately after end of exposure, and 15 and 60 minutes after the end of exposure Figure 9. Change in optical density and rate of change in optical density for GafChromic MD- 55 film as a function of time before, during and after exposure to 381 cgy with 6 MV X-rays.. 35 Figure 10. Termination of exposure is taken as the intercept of the two fitted lines: first line corresponding to data obtained during exposure and second line corresponding to data obtained after end of exposure Figure 11. Change in optical density can be calculated for any exposure by subtracting initial OD from final OD Figure 12. Schematic of setup for temperature dependency experiments Figure 13. Change in optical density for GafChromic MD-55 exposed to 381 cgy with 6 MV X-rays as a function of time Figure 14. Change in OD for five pieces of film, each exposed to 381 cgy with 6 MV X-rays (at the doserate of 285 cgy/min) Figure 15. Inferred dose using ΔOD measurements and calibration plot as a function of applied dose Figure 16. Change in optical density for a piece of GafChromic MD-55 film during several exposures applied approximately 5 minutes apart Figure 17. Optical density as a function of dose for a system utilizing one, two and four pieces of stacked film Figure 18. Change in optical density as a function of dose for doses delivered at 95 cgy/min, 286 cgy/min and 671 cgy/min Figure 19. Rate of change in optical density as given by the linear fit of data obtained during exposure as a function of applied dose, for doses delivered at 95 cgy/min, 286 cgy/min and 571 cgy/min Figure 20. Position of wavelength of maximum absorbance for GafChromic MD-55 as a function of irradiation/measurement temperature Figure 21. Change in OD for a given dose as a function of applied/measured temperature using both a constant spectral averaging window and a shifting spectral averaging window Figure 22. Schematic of experimental setup x

11 Figure 23. Emission of the light emitting diode used in experimental setup, as measured by spectrometer Figure 24. Schematic of layers in EBT film Figure 25. Change in absorbance of EBT film over a range of wavelengths before, immediately after, and at two time points post-exposure Figure 26. Optical density versus time for a single piece of EBT film; optical density versus time for five pieces of EBT film shown on a reduced time scale (inset) Figure 27. Wavelength of maximum absorbance for EBT film versus time during and after exposure to 9.52 Gy at 2.86 Gy/min with 6 MV X-rays Figure 28. Optical denisty of EBT film versus time for various spectral averaging windows Figure 29. Optical density for EBT and MD-55 films during and after exposure Figure 30. Percent increase in OD for EBT and MD-55 films after exposure, calculated with respect to OD at the end of exposure Figure 31. Percent increase in OD for EBT and MD-55 films within one hour after exposure.. 73 Figure 32. Optical density for EBT film exposed to 9.52 Gy, delivered with 6 MV at 0.95 Gy/min and 5.71 Gy/min Figure 33. Microscope images of monomer crystals within the sensitive media of MD-55 and EBT films Figure 34. Optical density versus time for a 50 cgy irradiation at 16 cgy/min Figure 35. The average sensitivity as a function of dose rate, for various doses Figure 36. Chemical formula of pentacoasa-10,12-dyinoic acid (PCDA), and lithium salt of PCDA (LiPCDA) Figure 37. Change in absorbance spectra for EBT film exposed to 1 Gy with 6 MV and 75 kvp beams Figure 38. Schematic of the modified phantom, with plastic water hoses on either side of the film and optical fibers Figure 39. Wavelength of maximum change in absorbance of commercial EBT films irradiated to 1 Gy as a function of measured temperature Figure 40. Values of wavelength of maximum absorbance for various doses delivered to commercial EBT films as a function of measured temperature Figure 41. Change in optical density for 1 Gy dose calculated for optical range of nm, and an optical range of 10 nm centered about wavelength of maximum absrobance, versus measured temperature xi

12 Figure 42. Temperature calculated using the position of wavelength of maximum absorbance versus measured temperature, shown with a line of best fit Figure 43. Change in optical density for films irradiated to 1 Gy using a fixed optical integration range of 630 to 640 nm, moving optical range of 10 nm about the peak of maximum absorbance, and as calculated using the peak of maximum absorbance and temperature-dependent correction factor Figure 44. Percent decrease in net OD for a 3 Gy dose, following different times in a desiccator at 50 ºC Figure 45. Spectral comparisons of absorbance of desiccated and normal unlaminated EBT film Figure 46. Absorbance of unlaminated EBT film after time in desiccator at 50 ºC Figure 47. Spectral comparisons of absorbance of desiccated, rehydrated, and normal unlaminated EBT film irradiated to 3 Gy Figure 48. Absorbance spectra of exposed unlaminated films using plate-like form of polymer, and the rehydrated form of hair-like polymer Figure cm 30 cm 4 cm phantom with the film insert Figure 50. Sample of time-dependent OD with time for a 1 Gy irradiation with a 75 kvp Therapax DXT 300 beam at 8 cgy/min for MD-55, HS, and EBT film Figure 51. Un-normalized change in OD for 1 Gy total dose for MD-55, HS and EBT films for irradiations delivered at various equivalent x-ray energies Figure 52. Change in OD per Gy for MD-55, HS and EBT, as a function of equivalent x-ray energy Figure 53. Increased sensitivity of HS and EBT films with respect to MD-55, for a dose of 1 Gy Figure 54. Change in absorbance of a single absorber versus time for different A parameters, keeping k and p parameters constant Figure 55. Change in absorbance of a single absorber versus time for different k parameters, keeping A and p parameters constant Figure 56. Change in absorbance of a single absorber versus time for different p parameters, keeping A and k parameters constant Figure 57. Experimental absorbance of MD-55 film at the main absorbance peak, and the model fit xii

13 Figure 58. Experimental absorbance of EBT film at the main absorbance peak, and the model fit Figure 59. Schematics of single and dual fibre optical dosimeter prototypes xiii

14 List of Abbreviations and Symbols ΔA change in absorbance ΔOD change in optical density ΔOD v change in visual density ΔV small volume ε(λ) extinction coefficient at wavelength λ λ wavelength λ max ρ 60 Co Cobalt-60 wavelength of maximum absorbance density A A(λ) AAPM b i c cgy D D Gy EBT E ev FWHM Gy HS HVL I I D I R I 0s I s ICRU IMRT absorbance absorbance at wavelength λ American Association of Physicists in Medicine polymer initiation constant per dose concentration centigray dose dose in Gy radiochromic film intended for External Beam Therapy energy electron-volt full width half maximum gray (J/kg) radiochromic film of High Sensitivity half-value layer intensity background intensity reference intensity initial intensity sample intensity International Commission of Radiation Units intensity modulated radiation therapy xiv

15 IGRT image guided radiation therapy ISP International Specialty Products k i kev kv p l LED LINAC LiPCDA polymer initiation constant kilo electron-volt peak kilovoltage length light emitting diode linear accelerator Lithium pentacosa-10,12-diynoate (lithium salt of PCDA) MD-55 radiochromic film intended for Medium Dose, size 5 5 MeV mega electron-volt MOSFET metal oxide semiconductor field-effect transistor MSDS Material Safety Data Sheet MV megavolt N 0 N m N ip N fp NTCP OD OD v OSL p i PCDA PDD PMMA QTH SAD SSD t TCP TG TLD initial number of monomer chains remaining number of monomer chains number of initiated polymer chains number of fully-formed polymer chains normal tissue complications probability optical density visual density (weighted by known response of human eye) optically stimulated luminescence polymerization kinetics constant pentacosa-12,12-diynoic acid percent depth dose poly-methyl methacralate quartz-tungsten-halogen source-to-axis distance source-to-surface distance time tumour control probability Task Group thermoluminescent dosimeter xv

16 TPR UV Z tissue-phantom ratio ultraviolet atomic number xvi

17 CHAPTER 1: INTRODUCTION 1

18 2 The work within this thesis describes a novel method for performing real-time dosimetry using fibre-optic read-out of radiochromic materials. The radiochromic materials are investigated for their applicability in clinical dosimetry measurements in vivo and in vitro. Their performance as a function of dose and time is modeled, with parameters linked to physical properties of the materials and the processes that occur during exposure to radiation. A system is thus established for evaluating radiochromic dosimeters for clinical dosimetry, whereby their performance can be at least in part be predicted by their physical properties. In this chapter, clinical rationale for the proposed real-time dosimeter is established by outlining the need for in vivo dosimetry and the inability of the dosimeters presently available on the market to meet that need. I. Ionizing Radiation in Cancer Treatment Ionizing radiation is encountered under many circumstances in medicine, and specifically in oncology. It is used to identify and locate the cancer, to target it, and to treat it, with approximately 50% of cancer patients receiving radiation therapy for management of their disease. * High energy photons (referred to as x-rays and gamma rays) are known to damage tissue. Although the exact details of tissue damage are still not fully understood, it is believed high energy photons induce ionization of important molecules within the cells, such as deoxyribonucleic acid. Ionization refers to removal of an electron from a molecule, making it unstable. These unstable molecules may react in a way that would prevent them from functioning properly, eventually leading to cell death. The source of radiation can be external or internal (known as brachytherapy), varying greatly in energy and intensity. Energy can vary from kev (1) gamma rays from decaying radionuclides in brachytherapy seeds, to 18 MV x-ray or 20 MeV electron beams from a linear accelerator (LINAC). The dose rate can be as low as Gy/h (1) (where Gy=1J/kg) for low dose rate brachytherapy, to as high as 6 Gy/min for LINAC treatments. The majority of the external treatments are divided into dose fractions delivered daily, five days a week, over several weeks, with only a small percentage of treatments delivered in a single large dose (known as stereotactic radiosurgery) from a 60 Co unit called GammaKnife or from a LINAC using a 6 MV x-ray beam. In recent years, treatments have become more conformal to the tumour due to implementation of Intensity Modulated Radiation Therapy (IMRT) and Image-Guided Radiation Therapy (IGRT). With the development of these new technologies, a trend has been evolving *

19 3 towards higher target absorbed dose values, fewer fractions, smaller treatment volumes, and steeper dose gradients. These developments need to be validated in terms of actual dose delivered. II. Radiation Dose Cell damage from ionizing radiation may result in several different outcomes, including repair of damage by the cell, cell death, and survival with mutation. (2) Depending on the type of damage and the tissue irradiated, the biological effect can take anywhere between a few hours (acute) to many years (late) to manifest. During radiation therapy, the dose and its distribution are important for the outcome of the treatment and prevention of further complications. A high enough dose has to be delivered to the tumour and affected organs to obtain high probability of tumour control, and a minimal dose should be delivered to healthy surrounding organs to limit probability of acute or late effects. (2) To maintain high probability of tumour control, the International Commission of Radiation Units (ICRU) recommends uniformity of tumour dose within +7/-5% of the total prescribed dose. (3) The upper limit exists because the dose prescribed is often limited by the dose delivered to the surrounding healthy organs during irradiation, which is dependent on the type of treatment delivery. Generally, conformal treatment allows for higher dose to be delivered to the tumour and for lower dose be delivered to the surrounding tissues, though the total volume of tissue irradiated may increase. On the other hand, the probability of cancer recurrence due to geometric miss may increase. The probability of biological effect taking place (whether it be tumour cell kill or normal tissue complications) versus dose is called a dose response curve. (2) The curves for tumour control probability (TCP) and normal tissue complication probability (NTCP) are often plotted as sigmoid relationships (Figure 1). That is, there is nearly no effect at first, then the probability rises sharply, and levels off to a plateau. Thus delivering a smaller dose to the tumour than that required for cure or control would sharply increase the probability of relapse. On the other hand, if the patient receives a dose to the tumour that is much higher than that prescribed, then it is also likely that the patient receives a higher dose to the surrounding normal tissue. Given the sigmoidal relationship between dose and effect, the risk of acute or late effects increases dramatically. Because of the conformality of modern treatments, describing the dose distribution by a single dose to the tumour and using the +7/-5% recommendation for guidance is an oversimplification. As the treatments become tailored to each patient s needs, the dose

20 TCP NTCP Probability 0.5-5% +5% Dose (Gy) Figure 1. Schematic of a typical relationship between tumour control probability (TCP) and normal tissue complication probability (NTCP) versus dose. The NTCP curve is based on two values (marked by X): doses at which 5% and 50% of patients develop complications when 2/3 of their liver is irradiated. (4) Dashed line represents a dose of 42.5 Gy, yielding 90% probability of tumour control, and 37.5% probability of normal tissue complications (liver failure). Increasing the tumour dose by 5% (dotted line) of the prescribed dose increases TCP by only 2.5%, but increases NTCP to 50%. On the other hand, delivering the same distribution with 5% lower dose (dotted line), decreases NTCP to 25%, but also decreases TCP to 82.5%, which may compromise treatment outcome. distributions become more diverse. Systematic errors larger than 5% in dose delivery (measured as entrance, exit dose, or combination of the two during treatment) to a small percent of patients (~ 1%) have been published. (5-7) These can be due to inadequacies in dose calculation algorithms, (8) setup errors, or a human error on behalf of the many individuals involved in the process of patient treatment. Although doses to the patient can be calculated or inferred from a

21 5 relative measurement, the American Association of Physicists in Medicine Task Group 40 recommended that clinics have access to an in vivo dosimetry system. (9) While radiation transport algorithms are constantly being improved in order to perform dose calculations, and phantoms become more complex to better represent human anatomy, conditions where dosimetry and simulation of humans remain a challenge still exist. Accurate assessment of the dose distribution in brachytherapy treatment and in regions near in homogeneities for external beam treatments is vital to rapid innovation and development of new radiation therapy technologies and techniques. Such an assessment can be done by performing in vivo dose measurements. However, performing such measurements under clinical conditions is challenging. III. Radiation Dosimetry A. Basic Interactions The interaction of ionizing radiation with matter results in a dose deposited within that matter, where dose is defined as the absorbed energy per mass (J/kg, or Gy). (10) In radiotherapy, the dose quoted is often dose to water, as most tissues within the body have similar radiological properties as water (common exceptions are lung tissue, bone and teeth). The ionizing radiation can be directly ionizing (charged particles) or indirectly ionizing (photons). (10) The photons, as they pass through the medium, are attenuated by the medium and scatter from their original path. The processes of attenuation are due to coherent scattering, photoelectric effect, Compton effect, pair/triplet production, or photonuclear interactions. In radiation therapy, the middle three are the interactions important for dose deposition, and the probability of any of these events happening when a photon beam passes through the medium depends on the energy of the photon, density, and the atomic number (Z) of the medium. Photoelectric effect is the predominant interaction for low-energy photons (below 100 kev). The probability of photon interacting in this manner in a medium of a given density increases with Z 3 of the medium. Here all of the photon s energy is transferred to one of the inner shell electrons of an atom. This electron then continues through the medium with a kinetic energy equal to the energy of the initial photon less the binding energy of the electron. In a Compton interaction, the photon interacts with a valence shell electron, transferring part of photon s energy to an electron and the two then continue at a given angle from each other from the point of interaction, such that the momentum is conserved. In this interaction, the photon does not transfer all of its energy to the electron. The probability of this interaction occurring in a medium of given density is

22 6 approximately independent of Z of the medium, and is predominant for photon energies around 1 mega-electron-volt (MeV). For high energy photons (over MeV) pair production can occur. In these interactions, as the photon interacts with the Coulomb field near the nucleus, the photon is absorbed giving rise to an electron and a positron. If the photon interacts with the field of the atomic electron, the atomic electron also acquires energy and escapes from the atom, thus yielding triplet production. The process of triplet production requires the energy of the photon to be greater than MeV. The probability of pair production for a medium of given density is roughly proportional to Z, while triplet production is independent of Z of the medium. (10) The electrons that result from these interactions of photons with the medium in turn travel through the medium themselves. The electron interacts with almost every atom in its path whose electric field is detectable. Most of these interactions transfer small fractions of the electron s energy, and thus the electron is often thought of as losing the energy gradually in a frictionlike process. Some of these energy losses result in a photon (Bremsstrahlung) being emitted when the electron changes direction due to an electric field from a nearby nucleus. The energy lost in this way does not contribute to the locally deposited dose. Electron interactions that do contribute to local dose will either excite the shell electrons of a nearby atom to a higher energy level, or ionize it. The rate of energy loss per distance decreases with increasing Z of the medium and increasing kinetic energy of the electron. (10) B. Standards and Protocols for Dosimetry Ionizing radiation dose can be measured well in standard conditions following set protocols (11-14). In most cases, using dosimetry equipment calibrated at a national standard laboratory (such as National Research Council of Canada) or traceable to such a calibration, and performing measurements under controlled conditions allows for accurate dosimetry in air and in water or plastic phantoms, with uncertainties below 1%. (12,15) However, in some cases, such as intravascular brachytherapy, absorbed dose standards may vary by as much as 10% between measurements and different Monte Carlo calculations. (16) When using the dosimetric gold standard, the ion chamber, some of the controlled conditions include a known temperature and pressure, type and energy of the beam, distance from the source and depth (if phantom is used). (12) Unfortunately, these parameters may not always be known to a high level of precision and accuracy in clinical conditions.

23 C. Estimation of Dose Delivered in Therapy 7 Doses under uncontrolled conditions, delivered during imaging and therapy, can be obtained in several different ways. They can be computed using various algorithms as is done in actual treatment planning, (17) with the algorithms constantly improving to include dose calculations for brachytherapy and imaging procedures, as well as external beam irradiations. (18-21) The dose can also be inferred from a relative dose measurement performed during the procedure (such as skin dose measurement), or measured directly at a point of interest. Because the points of interest on the patient may be inside the patient, such as at the tumour site, performing a direct measurement is often trickier than the other two approaches. IV. The Challenges of Dose Measurement in the Clinical Setting While real-time dosimetry may be useful for in vitro measurements at several points of time varying radiation fields, such as those in high dose rate brachytherapy or IMRT, the discussion of clinical applications here is kept to in vivo measurements. Predicting the three-dimensional cumulative dose distribution within a patient over the course of their treatment can be complex, given the variations arising from minor patient positioning errors, treatment beam fluctuations, motion during treatment, changes in anatomy as the patient loses weight or the tumour shrinks, and possibly other sources of error. Thus measurements of dose at or within a patient are often desired as part of a quality assurance, for investigative purposes of a new procedure, or for implementation of new technology protocol. Performing dose measurements inside patients is more complicated than skin dose measurements, and is not nearly as straightforward as phantom dosimetry which is often utilized in the clinic. A. Clinical Applications and Ideal Dosimeter A dosimeter that can accurately measure ionizing radiation dose in vivo under various clinical conditions may simplify and improve the current state of dosimetry. It can be used for quality assurance of both external beam and brachytherapy treatments. If an error in positioning, machine output, transcription or some other error occurs that results in a discrepancy of planned dose above a set threshold, treatment can be interrupted, and the discrepancy investigated. The dosimeter may also be used to track dose in an organ that moves in and out of the radiation field during respiration. In a brachytherapy procedure, the real-time dose rate measurement may be used as feedback to verify proper positioning of seeds. If the dose rate is higher or lower than

24 8 anticipated, the insertion of the next set of seeds may be adjusted to get the proper dose rate and dose at the point of measurement. For a dosimeter to be an appropriate option for in vivo measurements in most clinical scenarios, the overall construction and the dosimetric medium must satisfy several requirements, listed in Table 1. The requirements can be used as a set of guidelines for evaluation of a new dosimetric medium. The presence of the dosimeter must not perturb the tissue or the dose distribution. It must also be sufficiently small (1) to be able to resolve a sharp dose gradient in a radiation field. However, the size of the sensitive medium must be sufficiently large to yield good signal statistics for high precision measurements. In part, the signal statistics can be controlled if the dosimeter uses passive read-out method, as is discussed later in this Chapter. The dosimeter should have near water-equivalent composition (2), such that its response with change in energy is similar as that of water. (22) This allows for a single dosimeter to be used across all energies encountered both in external beam and brachytherapy treatments without performing a separate calibration. The dosimeter for quality assurance purposes should respond in real time (3), providing dose or dose rate estimate within a few seconds from beginning of irradiation so that treatment can be interrupted if the dose rate is outside of the value expected. The dosimeter also has to show a dose response that is independent of the dose rate used to deliver that dose (6). The dose estimate should have high dose resolution (5) in order to detect small variations (~1 cgy) in the dose delivered (daily fractions are often ~200 cgy, and total doses are Gy, depending on the treatment and fractionation pattern), and the response should be ideally linear with dose (4) for simple computation. Some of the other issues to be taken into consideration are environmental conditions (7), such as the temperature dependence of the dosimeter. The dose estimate should be independent of the temperature of the dosimeter, which can vary from near room temperature on skin surface to as high as ~38 C within the patient. Finally, the dosimeter must be non-toxic (8) and biocompatible, and would preferably be inexpensive enough to be disposable after each patient. A dosimeter, as described above, which can be used across a wide range of energies could be implemented in both external beam radiation therapy and brachytherapy, simplifying the dosimetric procedures to a single device. Accurate assessment of the dose given to the patient using a reliable dosimeter with a water-equivalent response would save time and money over using multiple dosimeters with a significant over- or under-response to low energy X-rays and inferring the dose via correction factors and calculations.

25 Table 1. List of criteria for in vivo point-based real-time dosimeter. 9 Criterion Criterion Comments # 1 small size (22) (<1 mm 3 ) Does not physically perturb tissue and effect delivered dose to surrounding tissue; can measure point dose at interfaces between tissues of varying densities and composition; used on high gradient radiation fields 2 near water-equivalent (22) (difference <10%) (response independent of energy) Does not alter dose distribution to tissue (tissue assumed to be similar to water) (22) ; own reading converted to dose delivered to water (and/or tissue); does not cause artefacts during lowenergy image guidance (for IGRT) 3 fast kinetics & stable response Both required for real-time readout of dose (read-out process does not induce false signal) 4 signal dose in cgy range (linear within 2%) Simplicity of conversion from measurements to dose; no need to track delivered dose to-date; simple function is acceptable in lieu of linearity 5 dose resolution (down to cgy) Measurements of doses down to a few cgy with relatively small errors (few %); suitable for IMRT 6 Dose-rate independence ( cgy/min) No need for prior knowledge of dose; simplicity of measurements throughout the body (no statistical difference) 7 Insensitive to environmental conditions (<2% variation over clinical Temperature, humidity and light insensitivity allows for easier incorporation and use within clinic temperature range of C) 8 non-toxic Requirements of dosimeter embodiments are relaxed when dosimeter within is non-toxic

26 B. Current in vivo Dosimeters 10 Several dosimeters currently available on the market are used for some in vivo dosimetry measurements. These are in vivo ion chambers, diodes and MOSFETs (metal oxide semiconductor field effect transistors). An ion chamber is classified as an absolute dosimeter: it can be used to measure the absorbed dose to its own sensitive volume without any calibration. (10) However, it often needs to be calibrated as knowing the exact measurement volume, and mass of air contained, is required for absolute dosimetry. The ion chamber measures the total ionization produced by electron interactions in air, the charge is collected by an electrode set at a high voltage, and this value can be related to dose. The mass of air contained must yield reasonable signal statistics, limiting in vivo ion chambers to dosimetry in bladder and rectum, too large for use in tissue. A silicon semiconductor diode is what is generally used as an in vivo radiation detector. When the diode is exposed to ionizing radiation, electron-hole pairs are created throughout the diode, and as they move through the diode, a measurable current is created. The amount of silicon required for measurement (known as the die) is mm 3, and the current related to deposited dose can be measured in real-time via the coaxial cable by the electrometer. (23) For in vivo dosimetry, the die is covered with material for protection and for proper buildup. The buildup is necessary for skin dose measurements to give the photons enough medium to interact with in order to create the electrons that will in turn interact with the silicon. (23) This buildup and the protective cover make the diodes rather bulky (up to 3 cm in length and ~7 mm in diameter), making it difficult or impossible to position into all tissues of the patient because of their size. Some diode configurations also suffer from angle dependence of their response (due to inhomogeneity of the buildup material and coaxial cable attached), and they have been shown to perturb the dose distribution directly behind the diode by as much as 30%, with the effect more pronounced at lower beam energies. (24) Read-out was also shown to vary with temperature, (24,25) complicating dosimetry further by the fact that the temperature of the diode may not always be known during its use. A MOSFET is a small silicon transistor. (26) A p-type silicon semiconductor sits on an insulating oxide layer, which separates it from a conducting metal band. (27) The p-type silicon has positive charges accumulated within, proportional to the negative voltage bias applied at the conducting metal band. The measurement of dose is related back to the threshold voltage (voltage required to allow current to flow through the semiconductor). (26,27) As the ionizing radiation travels through the MOSFET, charges generated within are trapped, causing the

27 11 threshold voltage to shift proportionally to deposited dose. (26,27) MOSFETs can be used as realtime dosimeters (27) and are much smaller than diodes (some are as small as 1 mm in diameter, known as micro-mosfets), decreasing beam attenuation and dose perturbation effects that are observed for diodes. (26) Their small size also allows for dosimetry in small beams, down to a few mm (4.4 mm) in beam diameter. (28) Having high Z components and not being water-equivalent, MOSFETs have large differences in calibration factors, (29) and require separate calibrations to be performed at different beam energies. (26) On the other hand, they show good agreement with the ion chamber for a given energy down to a depth of 34 cm. (30) They have also been shown to exhibit directional anisotropies in response because of the silicon substrate beneath the sensitive volume, (28,31) and, like diodes, are known to exhibit temperature dependence. (26) C. Optical Methods Energy independence of a dosimeter can in large part be met by staying clear of metallic components within the dosimeter. As such, there has been a considerable amount of effort over the last few decades to find a dosimeter based on optical characteristics of a radiation sensitive medium and fibre-optic readout. Among such radiation sensitive media are doped optical fibres, (32-34) plastic scintillators, (35-39) thermoluminescent dosimeters (TLDs), (34,40) optically stimulated luminescent (OSL) dosimeters, (34,41,42) and a fluorescing ruby. (43) The media can be subdivided into two categories: light emitters and light modifiers. Light emitting dosimeters (such as scintillators, TLDs and OSL media) produce signal that is proportional to the absorbed dose. Thus the number of photons and the signal statistics are dependent on dose, and is out of user s control. On the other hand, light modifying dosimeters alter some aspect of the interrogating light, the properties and the intensity of which is controlled by the user. This allows for higher precision measurements, because the number of photons can be increased if the noise is too high. The other major difference between the two types of optical dosimeters is that the light emitting media are reusable, whereas light modifying media have to be disposed off after a certain dose. Reusable dosimeters are often less expensive per use, but age and one must be careful to not assume the signal per dose remains constant as the total dose delivered is increased. Dosimeters that integrate dose to give a single signal at the end, such as light modifying media, have the ability to always keep the reading, and can be measured multiple times as the read-out is non-destructive. As they are also disposable, the need for disinfecting between patients is eliminated, simplifying their use, as well as reducing the risk of spreading infection.

28 12 Fluorescing rubies and scintillating fibres automatically emit light when exposed to ionizing radiation, whereas TLDs and OSL dosimeters must be stimulated by either heat or light, respectively, to obtain a light signal. These materials work by trapping electrons in higher energy states when they are exposed to ionizing radiation. When the electrons move down (either automatically or due to stimulation) to their ground state, photons corresponding to the energy difference between the two states are emitted. TLDs cannot be read out in real-time, as they require annealing after irradiation and a lengthy read-out processes for accurate dose estimate. (5) When one considers the high temperatures that TLDs must be heated to (at least 100 C, depending on the type of TLD), (28,40) it is hard to imagine how this would be done safely within a patient. Some fibre-based read-out schemes have been suggested, (40) but have not been implemented clinically. OSL and plastic scintillator dosimeters are promising, and some have been made to be nearly energy independent. (38) However, they continue to suffer from interferences such as fibre scintillation and Cerenkov radiation, where removal of the latter often requires accurate knowledge of pulse sequences and careful timing. (42,44,45) The alternative is to use scintillators that have high emission wavelength, such that the Cerenkov radiation (which drops off with 1/λ 3 ) from the fibre doesn t interfere much with the dose-related signal from the sensor. (37) However, the signal per given dose from such scintillators is generally decreased compare to the signal from scintillators with low emission wavelength, and thus measurements of dose are noisy. (37) While rubies fluoresce at high enough wavelength and long enough after the pulse such that the Cerenkov radiation is irrelevant to the measurement, they have a high Z and are not water equivalent. (43) Light modifying dosimeters, such as doped optical fibres and Fricke xylenol-orange solutions or gels create light-absorbing colour centres when exposed to ionizing radiation. This is done either via electron trapping, (46) or via formation of a complex with a dye, (47) respectively. Doped optical fibres are generally not water equivalent due to high Z (often Pb) components used as doping material. (33,34,46) A method for reducing Z by incorporating dopants such as Na, Mg, and Li has been proposed. (48) However, these optical fibres have not been implemented, likely because of reduced sensitivity compared to higher-z counterparts. Finally, while certain gels can be used as optical dosimeters, these are typically utilized in 3D dosimetry by making 3D phantoms out of the gel, (49-52) and no effort to incorporate them in fibre-optic dosimeter has been made; rather, they are used in post-exposure volumetric readout (e.g. MR).

29 V. Outline of Thesis 13 Another type of optical dosimeter makes use of what is known as a radiochromic medium. This type of material changes colour, or gets darker, upon exposure to ionizing radiation, and is in the category of light modifiers. Some radiochromic films are manufactured under the name of GafChromic (International Specialty Products, or ISP). These films contain one or two gelatin layers with organic monomers arranged in a small crystal or micelle-like structure suspended within them. (53-55) The monomers undergo polymerization when exposed to radiation. The absorbance spectrum of the resulting polymer systems is then related back to the absorbed dose. (55-57) Historically, these films are used for two-dimensional dose distribution measurements, (58,59) and the measurements are performed 3-24 hours (depending on the film) (54,55) after the end of irradiation to ensure stable readout. This is because the polymerization reaction is not instantaneous, and proceeds even after the source of ionizing radiation is removed. This, in turn, causes the absorbance to change with time, producing errors in dose estimate. (59-61) Despite the recommendation that these media be read out 3-24 hours after irradiation, radiochromic media are being investigated in this work for applicability in real-time patient dosimetry. They have some advantages, including a near water-equivalent organic composition, (54,55) and the ability to produce signal from a sub cubic millimetre volume. (55) They also absorb predominantly in the red region of the visible spectrum, where Cerenkov radiation does not interfere. Because the radiochromic material is a light modifier, the signal statistics can in part be controlled by the user, by increasing or decreasing the interrogation light. More importantly, if the performance of these systems during and after irradiation can be characterized and accounted for, they may provide real-time dose estimates with an acceptable error despite the above-mentioned issues. If these systems are understood, reverse engineering may be possible to create a radiochormic material that polymerizes faster and has appropriate sensitivity for a given application. In the present work, response of two films (GafChromic MD-55 and EBT) were assessed as a function of dose, time, dose rate, temperature, and energy and results are described in Chapters 2 to 6. Chapter 2 investigates in detail the possibility of using the radiochromic material, GafChromic MD-55, as a point-based dose measurement material in real-time. Although throughout the experiments described in this thesis the measurements were made immediately after the end of irradiation, the endpoint was chosen only because this is when the dose is known. It is easy to imagine how, once the relationship between optical density and dose is established,

30 14 optical density measurements can be made any time during irradiation. Thus, the dose can be estimated during irradiation as well, making it a true real-time dosimeter. The film was investigated for signal linearity, reproducibility, dose rate and temperature dependence. Chapter 3 compares the performance of GafChromic MD-55 with a medium from another film, GafChromic EBT. This chapter focuses on differences in sensitivity and linearity of MD-55 and EBT, and discusses the fundamental chemical and structural difference between the two monomer systems. Chapter 4 describes the investigation and quantification of the dose rate dependence of GafChromic EBT. Chapter 5 describes temperature and humidity investigations of GafChromic EBT (performed in collaboration with Dr. D.F. Lewis and Dr. S. Varma, researchers of ISP). Chapter 6 compares energy dependence between two sensitive media present in three films (MD-55, EBT and HS, where MD-55 and HS use the same formulation). A novel mathematical model of the response to dose with time both during and for short periods after the end of irradiation is also developed, with the preliminary results described in Chapter 7. The parameters of the model are based on physical properties and processes occurring during exposure to ionizing radiation and interrogation with read-out light. Ideally, this would allow for future engineering or selection of radiochromic media that meet the in vivo requirements, by working backwards from the desired response to radiation as predicted by the model to physical and chemical properties. The thesis concludes with a summary of current investigations and ideas for future work. 1. Suntharalingam N., Podgorsak E., Tolli H. Brachytherapy: Physical and Clinical Aspects. In: Podgorsak E., ed. Radiation Oncology Physics: A Handbook for Teachers and Students. Vienna: International Atomic Energy Agency, p , Suntharalingam N., Podgorsak E., Hendry J.H. Basic Radiobilogy. In: Podgorsak E., ed. Radiation Oncology Physics: A Handbook for Teachers and Students. Vienna: International Atomic Energy Agency, p , ICRU 50. Prescribing, Recording, and Reporting Photon Beam Therapy. Bethesda: International Commission on Radiation Units and Measurements, 1993.

31 15 4. Burman C., Kutcher G., Emami B., Goitein M. Fitting of normal tissue tolerance data to an analytic function. International Journal of Radiation Oncology Biology Physics 21: , Essers M., Mijnheer B.J. In vivo dosimetry during external photon beam radiotherapy. International Journal of Radiation Oncology Biology Physics 43: , Feldman A., Edwarads F.M. The routine use of personal patient dosimeters is of little value in deteting therapeutic misadministrations; Point/Counterpoint. Medical Physics 28: 295-7, Noel A., Aletti P., Bey P., Malissard L. Detection of errors in individual patients in radiotherapy by systematic in vivo dosimetry. Radiotherapy and Oncology 34: , Papanikolaou N, Battista JJ, Boyer AL, Kappas C, Klein E, Mackie TR, Sharpe M, Van Dyk J. Tissue Inhomogeneity Corrections for Megavoltage Photon Beams. Medical Physics Publishing, Kutcher G., Coia L., Gillin M., Hanson G.K., Leibel S., Morton R.J., Palta J.R., Purdy J.A., Reinstein L.E., Svensson G.K., Weller M., Wingfield L. Comprehensive QA for radiation oncology: Report of AAPM Radiation Therapy Committee Task Group 40. Medical Physics 21: , Attix F.H. Introduction to Radiological Physics and Radiation Dosimetry. Toronto: John Wiley & Sons, Schulz R.J., Almond P.R., Cunningham J.R., Holt J.G., Loevinger R., Suntharalingam N., Wright K.A., Nath R., Lempert G.D. A protocol for the determination of absorbed dose from high-energy photon and electron beams. Medical Physics 10: , Almond P.R., Biggs P.J., Coursey B.M., Hanson W.F., Huq M.S., Nath R., Rogers D.W.O. AAPM's TG-51 protocol for clinical reference dosimetry of high-energy photon and electron beams. Medical Physics 26: , 1999.

32 Ma C.M., Coffey C.W., DeWerd L.A., Liu C., Nath R., Seltzer S.M., Seuntjens J.P. AAPM protocol for kv x-ray beam dosimetry in radiotherapy and radiobiology. Medical Physics 28: , Khan F.M., Doppke K., Hogstrom K.R., Kutcher G.J., Nath R., Prasad S.G., Purdy J.A., Rozenfeld M., Werner B.L. Clinical electron-beam dosimetry: Report of AAPM Radiation Therapy Committee Task Group No. 25. Medical Physics 18: , Shortt K.R., Ross C.K., Schneider M., Hohlfeld K., Roos M., Perroche A-M. A comparison of absrobed dose standards for high-energy X-rays. Physics in Medicine & Biology 38: , Chiu-Tsao S.T., Schaart D.R., Soares C.G., Nath R. Dose calculation formalisms and consensus dosimetry parameters for intravascular brachytherapy dosimetry: Recommendations of the AAPM Therapy Physics Committee Task Group No Medical Physics 34: , Knöös T., Wieslander E., Cozzi L., Brink C., Fogliata A., Albers D., Nyström H., Lassen S. Comparison of dose calculation algorithms for treatment planning in external photon beam therapy for clinical situations. Physics in Medicine and Biology 51: , Storchi P., Woudstra E. Calculation of the absorbed dose distribution due to irregularly shaped photon beams using pencil beam kernels derived from basic beam data. Physics in Medicine and Biology 41: , Aspradakis M.M., Redpath A.T. A technique for the fast calculation of three-dimensional photon dose distributions using the superposition model. Physics in Medicine and Biology 42: , Carlsson A.K., Ahnesjö A. Point kernels and superposition methods for scatter dose calculations in brachytherapy. Physics in Medicine and Biology 45: , Inanc F. A CT image based deterministic approach to dosimetry and radiography simulations. Physics in Medicine and Biology 47: , 2002.

33 Dutreix A., Bridier A. Dosimetry for External Beams of Photon and Electron Radiation. In: K.R.Kase, B.E.Bjarngard, F.H.Attix, eds. The Dosimetry of Ionizing Radiation. Toronto: Academic Press, Inc., Yorke E., Alecu R., Ding L., Fontenla D., Kalend A., Kaurin D., Masterson-McGary M.E., Marinello G., Matzen T., Saini A., Shi J., Simon W., Zhu T.C., Zhu X.R. Diode in vivo dosimetry for patients receiving external beam radiation therapy. Medical Physics Publishing, Marre D., Marinello G. Comparison of p-type commercial electron diodes for in vivo dosimetry. Medical Physics 31: 50-6, Saini A.S., Zhu T.C. Temperature dependence of commerically available diode detectors. Medical Physics 29: , Izewska J., Rajan G. Radiation Dosimeters. In: Podgorsak E., ed. Radiation Oncology Physics: A Handbook for Teachers and Students. Vienna: International Atomic Energy Agency, p , Gladstone D.J., Chin L.M. Automated data collection and analysis system for MOSFET radiation detectors. Medical Physics 18: 542-8, Francescon P., Cora S., Cavedon C., Scalchi P., Reccanello S., Colombo F. Use of a new type of radiochromic film, a new parallel-plate micro-chamber, MOSFETs, and TLD 800 microcubes in the dosimetry of small beams. Medical Physics 25: , Lavallée M., Gingras L., Beaulieu L. Energy and integrated dose dependence of MOSFET dosimeter sensitivity for irradiation energies between 30 kv and 60 Co. Medical Physics 33: , Chuang C.F., Verhey L.J., Xia P. Investigation of the use of MOSFET for clinical IMRT dosimetric verification. Medical Physics 29: , Scalchi P., Francescon P., Rajaguru P. Characterization of a new MOSFET detector configuration for in vivo skin dosimetry. Medical Physics 32: , Bueker H., Haesing F.W. Fiber-optic radiation sensors. SPIE 2425: , 1994.

34 33. Gripp S., Haesing F.W., Bueker H., Schmitt G. Clinical in vivo dosimetry using optical fibers. Radiation Oncology Investigations 6: 142-9, Huston A.L., Justus B.L., Falkenstein P.L., Miller R.W., Ning H., Altemus R. Remote optical fiber dosimetry. Nuclear Instrumets and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms 184: 55-67, Becks K.H., Drees J., Goldmann K., Gregor I.M., Heintz M. A multichannel dosimeter based on scintillating fibers for medical applications. Nuclear Instrumets and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 454: , Arnfield M.R., Gaballa H.E., Zwicker R.D, Islam Q., Schmidt-Ullrich R. Optimum plastic scintillator and optical fiber combinatin for brachytherapy dosimetry. International Journal of Radiation Oncology, Biology, Physics 32[Supplement 1], Archambault L., Arsenault J., Gingras L., Beddar A.S., Roy R., Beaulieu L. Plastic scintillation dosimetry: Optimal selection of scintillating fibers and scintillators. Medical Physics 32: , Beddar A.S., Briere T.M., Mourtada F.A., Vassiliev O.N., Liu H.H., Mohan R. Monte Carlo calculations of the absrobed dose and energy dependence of plastic scintillators. Medical Physics 32: , Petric M.P., Robar J.L., Clark B.G. Development and characterization of a tissue equivalent plastic scintillator based dosimetry system. Medical Physics 33: , Huston A.L., Justus B.L. All-optical, rapid readout, fiber-coupled thermoluminescent dosimeter system. [Patent US ] United States. 41. Aaki F., Ishidoya T., Ikegami T., Moribe N., Yamashita Y. Application of a radiophotoluminescent glass plate dosimeter for small field dosimetry. Medical Physics 32: , Gaza R., McKeever S.W.S., Akselrod M.S. Near-real-time radiotherapy dosimetry using optically stimulated luminescence of Al2O3:C: Mathematical models and preliminary results. Medical Physics 32: , 2005.

35 43. Jordan K.J. Evaluation of ruby as a fluorescent sensor for optical fiber-based radiation dosimetry. SPIE 2705: 170-8, Archambault L., Beddar A.S., Gingras L., Roy R., Beaulieu L. Measurement accuracy and Cerenkov removal for high performance, high spatial resolution scintillation dosimetry. Medical Physics 33: , Clift M.A., Johnston P.N., Webb D.V. A temporal method of avoiding the Cerenkov radiation generated in organic scintillator dosimeters by pulsed mega-voltage electron and photon beams. Physics in Medicine and Biology 47: , Bishay A. Radiation induced color centers inmulticomponent glasses. Journal of Non- Crystalline Solids 3: , Gupta B.L., Narayan G.R. G(Fe3+) values in the FBX gosimeter. Physics in Medicine and Biology 30: , Hasing F.W., Pfeiffer F., Buker H. Multipass cavity sensor for measuring a tissueequivalent radiation dose. Forschungszentrum Julich GmbH. [Patent US ] Bero M.A. Development of a three-dimensional radiation dosimetry system. Medical Physics 29(12), Oldham M., Siewerdsen J.H., Kumar S., Wong J., Jaffray D.A. Optical-CT gel dosiemtry I: Basic investigations. Medical Physics 30: , Oldham M., Siewerdsen J.H., Shetty A., Jaffray D.A. High resolution gel-dosimetry by optical-ct and MR scanning. Medical Physics 28: , Wuu C.S., Schiff P., Maryanski M.J., Liu T., Borzillary S., Weinberger J. Dosimetry study of Re-188 liquid balloon for intravascular brachytherapy using polymer gel dosimeters and laser-beam optical CT scanner. Medical Physics 30: 132-7, Klassen N.V., van der Zwan L., Cygler J. GafChromic MD-55: Investigated as a precision dosimeter. Medical Physics 24: , Todorovic M., Fischer M., Cremers F., Thom E., Schmidt R. Evaluation of GafChromic EBT prototype B for external beam dose verification. Medical Physics 33: , 2006.

36 McLaughlin W.L., Yun-Dong Chen, Soares C.G., Miller A., Van Dyk G., Lewis D.F. Sensitometry of the response of a new radiochromic film dosimeter to gamma radiation and electron beams. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 302: , McLaughlin W.L., Puhl J.M., Al-Sheikhly M., Christou C.A., Miller A., Kovacs A., Wojnarovits L., Lewis D.F. Novel radiochromic films for clinical dosimetry. Radiation Protection and Dosimetry 66: 263-8, McLaughlin W.L., Al-Sheikhly M., Lewis D.F., Kovacs A., Wojnarovits L. A radiochromic solid-state polymerization reaction. In: Irradiation of Polymers: Fundamentals and Technological Applications. Washington: American Chemical Society, p , Ramani R., Lightstone A.W., Mason D.L.D., O'Brien P.F. The use of radiochromic film in treatment verification of dynamic stereotactic radiosurgery. Medical Physics 21: , Mack A., Mack G., Weltz D., Scheib S.G., Böttcher H.D., Seifert V. High precision film dosimetry with GafChromic films for quality assurance especially when using small fields. Medical Physics 30: , Ali I., Costescu C., Vicic M., Dempsey J.F., Williamson J.F. Dependence of radiochromic film optical density post-exposure kinetics on dose and dose fractionation. Medical Physics 30: , Chu R.D.H., Van Dyk G., Lewis D.F., O'Hara K.P.J., Buckland B.W., Dinelle F. GafChromic dosimetry media: A new high dose, thin film routine dosimeter and dose mapping tool. Radiation Physics and Chemistry 35: , 1990.

37 CHAPTER 2: REAL-TIME RESPONSE OF GAFCHROMIC MD-55 FILM TO IONIZING RADIATION Portions of the following have been published as Suitability of radiochromic medium for realtime optical measurements of ionizing radiation dose by Alexandra Rink, I. Alex Vitkin, and David A. Jaffray in Medical Physics 32(4), p (2005) 21

38 22 I. Introduction The goal of these investigations is to develop a dosimeter that is economically and logistically acceptable (low-cost, disposable, reusable, and sterilizable). To meet these requirements, a water equivalent dosimeter which undergoes an immediate change in optical properties upon exposure to ionizing radiation is proposed. The difference in a particular quantitative optical property is to be measured via optical fibers, and ionizing radiation dose is to be inferred through a calibration model. In the initial embodiment of the device, the radiation sensitive material present in GafChromic MD-55 radiochromic film was investigated for suitability in the application of external beam patient dosimetry. A list of requirements in choosing an in vivo dosimeter against which GafChromic MD-55 film is investigated is given in Table 2. Review of GafChromic MD-55 To better understand the results presented in this paper, the reader is provided with a review of literature and explanation of solid-state polymerization. An understanding of the process that forms the basis for radiochromic dosimetry is required if the real-time dosimetry system is to be quantified and optimized. The time course of the energy transfer and subsequent processes that lead to changes in optical density are also important for rational design. General Experience Radiochromic films have been used for nearly 30 years in the field of dosimetry. (1) Commercially available radiochromic dosimeters are manufactured by International Specialty Products (ISP), and some are sold under the product name of GafChromic MD-55. A broad assessment of its characteristics suggest that it is a good candidate for the proposed point-based dosimeter: the sensitive medium from GafChromic MD-55 film can be packaged as a small volume placed at the tip of an optical fiber (closed system to minimize any interference from the tissue, such as humidity); it has response characteristics within 5% of water and striated muscle for photons of energy in the range of MeV, and electrons in range of MeV. (2) Upon exposure to heat, ultraviolet (UV) light, and high-energy photons and electrons, the monomers polymerize to provide an absorbance spectrum with two peaks (675 and 615 nm), creating a polymer with a blue tint. (2-4) The signal linearity requirement appears to have also been met, since the change in absorbance is a linear function of the absorbed dose, (5) although the dynamic range of this function depends on the wavelength at which the measurements are

39 Table 2. Evaluation criteria for in vivo point-based real-time dosimeter. 23 Criterion Criterion Comments # 1 small size (6) (<1 mm 3 ) Does not physically perturb tissue and effect delivered dose to surrounding tissue; can measure point dose at interfaces between tissues of varying densities and composition; used on steep part of dose-depth curve or in the penumbra region; no build-up required for measurement 2 near water-equivalent (6) (difference <10%) (response independent of energy) Does not alter dose distribution to tissue (tissue assumed to be similar to water) (6) ; own reading converted to dose delivered to water (and/or tissue); does not cause artefacts during low-energy image guidance (for IGRT) 3 fast kinetics & stable response Both required for real-time readout of dose (interrogation process does not induce false signal) 4 signal dose in cgy range (linear within 2%) Simplicity of conversion from measurements to dose; no need to track delivered dose to-date; simple function is acceptable in lieu of linearity 5 dose resolution (down to cgy) Measurements of doses down to a few cgy with relatively small errors (few %); suitable for IMRT 6 dose-rate independence ( cgy/min) No need for prior knowledge of dose; simplicity of measurements throughout the body (no statistical difference using α = 0.05) 7 insensitive to environmental conditions Temperature, humidity and light insensitivity allows for easier incorporation and use within clinic (<2% variation over clinical temperature range of C) 8 non-toxic Requirements of dosimeter embodiments are relaxed when medium contained within is non-toxic

40 obtained. (7-11) 24 The film has been reported to resolve dose to 1.5 Gy with a precision of 5% or better, using the 671 nm absorption peak, and this resolution can be further increased by increasing the thickness of sensitive layer. (12) The requirement of real-time readout appears to be a significant impediment to use of GafChromic MD-55 film. Time frames of hours after exposure are recommended. Additional investigations are required to resolve this issue. Less than 5% difference in net change in absorbance of GafChromic MD-55 film exposed to 10 Gy at dose rates of Gy/min is expected. However, validity of the measurements done with this film has been questioned for low dose-rate brachytherapy. Ali and colleagues reported in 2003 the kinetics of film darkening as function of post-exposure time depends on the total dose, with the development being faster at the lower doses. (13) These findings are a concern for real-time dosimetry and require further investigation. While the focus of this study is to apply the dosimeter in the context of external beam radiotherapy, where doserates are typically greater than those in brachytherapy, a range of doses and dose-rates, over which post-exposure development from the first few fractions of the treatment does not introduce error in the absorbance reading and final dose estimate, should be clearly defined. McLaughlin et al. [1996] reported that propagation of the polymerization is complete within 2 ms of a single 20 Gy 50 ns pulse. (4) It is unclear, however, if the polymerization occurred mostly due to ionizing radiation or heat. There literature describes a continuous increase in absorbance even after irradiation is complete, (14,15) with the absorbance being a function of a logarithm of elapsed time. (16) Hence, it has generally been recommended to perform the measurements 24 h (2) to 48 h later (16) by both researchers and manufacturers. * Measurements are further complicated by the shift in wavelength of maximum absorbance (λ max ) to lower wavelengths as dose increases. (2,7,14,15) GafChromic MD-55 film is stable during storage or short exposures to ambient light, (17) satisfying part of seventh requirement of in vivo dosimeter. However, the temperature dependence of absorbance of GafChromic MD-55 is complicated, and humidity and pressure dependence poorly documented. Increase in temperature reported to correspond to a decrease in absorbance and a peak shift to lower λ (λ max = nm at 18.6 C, 673 nm at 28.0 C for 6.9 Gy), (2,16) with the latter effect being reversible if temperature fluctuations occur during International Specialty Products (ISP) product information.

41 25 measurement, not but irreversible if the temperature was varied during irradiation. (2) Others report an increase in absorbance with an increase in temperature. (14,18) This discrepancy is likely due to a choice of wavelength for absorbance measurements, as the λ max depends on temperature, and also due to a range of temperatures sampled. It has been shown that a He-Ne laser operating as low as 0.1 mw will cause an increase in absorbance of GafChromic MD-55 in five minutes, with this effect being stronger for films exposed to smaller doses. (19) For this reason, the absorbance measurements should be performed using low optical powers to prevent polymerization due to the heat produced by the light. Above 60 C, the colour of the film changes from blue to red, as the crystals melt. (4) Chu et al., [1990] reported a dependence of GafChromic MD-55 absorbance on the relative humidity, with the increase from 35% to 100% humidity resulting in a decrease in sensitivity by approximately 14% when irradiated to 5-40 kgy. (14) No similar data has been published to date for doses of relevance to external radiation therapy. The crystal suspension used in GafChromic MD-55 film is hazardous if ingested, inhaled or absorbed through skin, although it would not be present in a substantial enough amount within the point-based dosimeter to do damage to the patient. Overall, GafChromic MD-55 is a stable, well-performing dosimeter medium, that appears to meet several of the requirements for the proposed in vivo dosimeter system. However, most of the published literature applies to measurements performed hours postexposure, and may not hold true during or immediately after exposure. Additional investigations are required to evaluate the performance of GafChromic MD-55 for real-time dose measurements. These investigations are performed using the novel optical fiber-based system described below. Solid-state Polymerization of Diacetylenes The active component of GafChromic MD-55 radiochromic film is a double-substituted diacetylene monomer with one polar end, organized into a crystal (Figure 2a). (2,16) The packing of the monomers within the crystal lattice depends on the type and size of the side groups (R 1 and R 2 ). (20,21) Generally, for polymerization to occur, the monomers should be packed such that the triple bonds of adjacent monomers are within an approximate distance of 0.4 nm or less. (22) It has been shown that a diacetylene monomer similar to that in GafChromic MD-55, upon MSDS, prepared by ISP, 01/11/2002. Dr. David Lewis, ISP ( ) (personal communication).

42 26 a b c d R1 R1 C C C C R1 C C C C R2 R1 C C C C R2 R1 C C C C R2 R1 C C C C C C C C R2 R2 R2 R1 C C R1 C C C C R2 C C R2 R1 C C R1 C C C C R2 C C R2 R1 C C C C R1 C C R1 C C R2 C C R2 C C a b c d R2 Figure 2. Structures of (a) diacetylene monomers, upon exposure to ionizing radiation, polymerizes into (b) butatriene structure polymer; as the polymer chain grows, it rearranges via (c) an intermediate between butatriene structure and acetylene structure, into (d) acetylene structure polymer. exposure to heat, X-rays, γ-rays or photons in the UV region, polymerizes into a butatriene structure polymer (Figure 2b). (23) For this particular monomer, this structure is stable in short chains (i.e. n 6). Once the polymer chains propagate longer (n>6), polydiacetylene undergoes reformation via an intermediate (Figure 2c) into an acetylene structure (Figure 2d) with a carbene (carbon atom with two unpaired electrons) at each end of the chain. The acetylene structure is of a lower energy configuration, (23) and any additional monomer attach to the chain will do so keeping the pattern of double and triple bonds along the chain the same. The solid-state polymerization of GafChromic MD-55 and other similar diacetylene monomer crystals is topotactic, where the reaction proceeds along one axis of the crystal, but not in the perpendicular axes. (22) The individual chains, once randomly initialized (24) by energy transferred from ionizing particles or scattered photons, grow independently of each other. (22) As the polymer increases in units, the absorbance of the polymer-monomer mixture increases due to an increase in the double-bond concentration between the units. (25) The shape of the absorbance spectrum is very similar (two major peaks, with larger peak at higher wavelengths) to that of

43 27 GafChromic MD-55 film and is said to be typical of a guest diacetylene polymer in a monomer crystal. (23,25) The structure of a diacetylene polymer is often slightly different than that of the crystalline monomer chain due to tilting of each unit with polymerization. (22) Initially, the crystal structure of the polymer is controlled by the crystal structure of the surrounding monomer. (22) Once the local polymer concentration increases and the crystal structure of the monomer can no longer produce enough strain to control the polymer structure, the polymer chains will reorganize into a lower energy arrangement. A transition in the structure of the diacetylene backbone is believed to be accompanied by a shift in the absorption peaks, and by a change in length of polymer chain. (26) This explains the shift of the absorption peak with increased dose. (2,7,14,15) In the case of GafChromic MD-55, the polymer structure is shorter, and as the chain grows, it contracts. Hence, the space between the last monomer unit in a chain and the next monomer unit increases as the polymerization proceeds. In order for polymerization reaction to proceed to the end of monomer chain, any mismatch of monomer and polymer must not lead to large spatial separation between the two phases. (20) However, with increasing concentration of polymers, the crystal structure required for polymerization is disturbed, and the number of units in a polymer chain per amount of energy deposited decreases as polymerization proceeds. Generally, the carbenes on both ends of the chain will continue to react until they reach an impurity that inhibits or significantly slows further polymerization, (24) or they terminate polymerization upon interaction with another carbene. (27) Since GafChromic MD-55 is ~ 99% free of chemical impurities, the large separation between the polymer chain and the adjacent free monomer acts as an impurity. It is expected that most of the energy is transferred to surrounding media (monomer crystals) within 10-3 seconds of a passage of high-energy electron. (28) The transferred energy would initiate polymerization within the monomer crystal. For a diacetylene monomer crystal similar to that in GafChromic MD-55, it has been reported that the butatriene-to-acetylene phase-transition occurs over a period of 4 hours at the temperature of 100 K. (23) It is reasonable to believe that the rate of phase-transition at ~ 23 C is much higher, and possibly nearly instantaneous. (24) Although carbenes are unstable and highly reactive species (27) and carbene reactions generally have high reaction rates, the reactivity of diacetylene monomers is a function of their spatial arrangement within the lattice. (20,21) Since the distance between monomers grows Dr. David Lewis, ISP ( ) (personal communication).

44 28 as the polymerization proceeds due to chain contraction, the reactivity of each new monomer to the polymer chain decreases; therefore, the kinetics of such a diacetylene polymerization within monomer crystals is not constant. At some point during the polymerization, the energy of the carbenes on the end of the chain is not sufficient to overcome the energy barrier created by the large distance to the next free monomer, and the probability of adding another monomer to the polymer drastically decreases. Nonetheless, a large percentage of an increase in absorbance would occur during exposure, and is referred from now on as intra-exposure growth. The polymerization reaction appears as if the crystal is heavily doped. (4) Carbenes that continue to react after the exposure has completed, contribute less than 20% to the overall change in absorbance, producing post-exposure, or inter-exposure, development. This inter-exposure growth in change in OD (ΔOD) is asymptotic in nature. (16) As shown in Figure 3, the rate of intra-exposure ΔOD is proportional to the dose-rate, and the net change in ΔOD is therefore proportional to absorbed dose. This model should hold true assuming the dose is small and the dose-rate is large, such that the polymer concentration is not sufficiently high enough to decrease polymerization kinetics for the subsequent dose fractions. For larger delivered doses, or for doses delivered at low dose-rates, the inter-exposure development from each deposition of dose will introduce errors to dose estimates due to reasons m 2 Optical Density m dd / dt 1 ΔD m 0 T Time Figure 3. A model of optical density of GafChromic MD-55 versus time before, during and after exposure. The irradiation is applied for a period of time T. The change in optical density during exposure is proportional to applied dose. m i are the slopes of lines, T is the irradiation time, and ΔD is the change in dose. International Specialty Products (ISP) product information.

45 29 discussed above. To test the model shown in Figure 3, a system and method that enables measurements during exposure is needed. The use of optical fibers for remote dosimetry has been previously reported on, (29,30) including systems that work by monitoring the degree of radiation-induced absorption within the fiber. (31,32) We report a novel fiber-based optical readout configuration that enables a medium, such as GafChromic MD-55 film, to be evaluated for suitability in a real-time point-based dose measurement system. The described dosimetry system, including optical readout configuration and method, exploits the intra-exposure changes in OD within radiochromic medium to estimate radiation doses in real-time, where dose measurements would generally be performed 48 hours after exposure using conventional methods. The system contains no metallic components near point of measurement. This approach permits water-equivalent dosimetry in small sensitive volumes over a large range of energies, dose rates, and doses; such a system can be used as a platform for testing other dosimetric media. The system is used to evaluate the sensitive medium employed in GafChromic MD-55 with respect to the eight requirements of the ideal dosimeter described above. II. Methods and Materials The optical dosimetry system is schematically illustrated in Figure 4. Since the main focus of this paper is the real-time behavior of GafChromic MD-55 film, no considerable amount of time was spent in perfecting the design of the optical system. A Roithner Lasertechnik (Vienna, Austria) light-emitting diode (LED), with peak emission wavelength of 670 nm (Figure 5), was chosen as the light source to provide the greatest sensitivity (10,11) and eliminate any unnecessary light that may heat the sample, thus altering measurements. (19) It was connected to a multimode optical fiber (600/630 core/cladding diameters in μm, Thorlabs, Newton, NJ) leading to an OZ Optics Ltd. (Carp, Ontario, Canada) non-polarizing 90:10 (sample:reference at 680 nm) beam splitter. The sample beam traveled through 17 m of fused silica fiber (600/630 core/cladding) to delivery fiber, which illuminated a spot ~650 μm in diameter of a 1cm 1cm piece of GafChromic MD-55 film contained within a cylindrical Solid Water phantom (Figure 6a-c and Figure 8). The dimensions of fused silica delivery and collection fibers are shown in Figure 8. The phantom was designed to allow easy replacement of film, to provide electron equilibrium at the point of measurement, located at the centre of the phantom, and to allow for

46 30 easy calculation of dose delivered to the point of measurement by using peak scatter factor for the linear accelerator on which exposures were performed. (33) The air gaps within the film holder and phantom were kept to a minimum: sufficient space to place the film, and small cylindrical spaces to allow passage of interrogating light to the film as well as collection of most of the light transmitted. The scattering was assumed to be negligible, and the fraction of light reflected was assumed to be independent of dose. A recent study by Fusi et al. [2004] confirmed that in the nm region, the absorption coefficient of GafChromic MD-55 film is larger than the scattering coefficient. On the other hand, they showed that total reflectance decreased from ~14% to ~9% over 0-4 Gy range. (34) However, no corrections were performed on our data in light of this information. Film in Solid H 2 O Phantom Detector Assembly Radiation Barrier Light Source Beam Splitter 90% 10% Signal Reference Detector Measurement and Processing Assembly Processor Figure 4. Schematic of experimental setup. Detector is an Ocean Optics Inc. SD2000 dualchannel spectrophotometer. Processor is a computer with Ocean Optics Inc. IOBase32 and Matlab 6.1.

47 31 Figure 5. Emission spectrum of the LED as detected by the spectrophotometer. Light transmitted through the film returned via another 17 m (same as above) of optical fiber into the signal channel of a spectrophotometer (Ocean Optics Inc., Dunedin, FL, USA; SD2000 dual-channel, 12-bit Analog-to-Digital Converter). The 10% reference beam was attenuated using 2.00 OD neutral density filter (Melles Griot, Carlsbad, CA, USA), and was monitored on the second spectrophotometer channel to measure fluctuations in the output light during evaluation of effect of light on OD of film (described in detail below). The spectrometer software used was Ocean Optics Inc. OOIBase32 (Dunedin, FL, USA), and data was processed with Matlab v6.1 (Natick, MA, USA). Briefly, the system operated with 30 ma power supply driving light output from the LED. Spectra were collected starting 5-10 seconds prior to the commencement of exposure. The spectrometer integration time ranged from ms, depending on the study performed. For integration time <500 ms, the spectra were captured no faster than 2 Hz. The change in absorbance (ΔA) for each wavelength (λ) was calculated using I os ( λ) I D ( λ) Δ A( λ) log10 ( ) (1) I ( λ) I ( λ) where I os is the average intensity of five spectra (taken before the beam is turned on), I s is the s D

48 a) b) 32 c) d a 2.5 mm 3.0 cm 2.5 mm 1.0 cm 1.0 cm + r = 1.0 cm b c 1.25 cm cm + mm a b d c 5.0 mm 1.0 cm 5.0 mm e f c d mm f 2.5 mm 1.0 cm e 1.5 cm 1.0 cm 3.0 cm Figure 6(a-c). Solid Water phantom (a) assembled, (b) disassembled, (c) schematic.

49 mm DELIVERY FIBER opening = mm buffer = mm cladding = mm core = mm NA ~ mm 1.78 mm 5.0 mm FILM ~0.25 mm 1.55 mm 1.59 mm COLLECTION FIBER opening = mm cladding =1.550 mm core = mm NA ~ 0.39 Figure 7. Schematic of cross-section of film holder in Solid Water phantom. The light emitted from delivery fibers illuminates a spot on the GafChromic MD-55 film ~650 μm in diameter. sample light intensity at some point in time, and I D is the dark signal. I D is defined as the average of the signal obtained over a period of 15 minutes without a light source. Figure 8 shows the change in absorbance of a single piece of GafChromic MD-55 prior to exposure, immediately after exposure to 381 cgy at 286 cgy/min, and at several intervals after the completion of exposure. The spectral window of interest, or range for optical density calculation, is a 10 nm window below the main peak ( nm). This window was chosen to minimize the errors due to shifting of wavelength of maximum absorbance with dose, (2,7,14,15) yet still provide a high sensitivity. (11) The optical density (OD) was then defined as 1 n 1 ΔA( λi ) + ΔA( λi+ 1) OD ( ) ( λ 1) i= 1 i λi+ λ λ 2 1 n where λ 1 to λ n are wavelengths that span the window of interest in the spectrum. The range of (2)

50 MV 381 cgy Prior to Exposure Immediately After Exposure 15 min After Exposure 1 hr After Exposure ΔA Spectral Averaging Window Wavelength (nm) Figure 8. Change in absorbance of GafChromic MD-55 film at various wavelengths plotted before exposure, immediately after end of exposure, and 15 and 60 minutes after the end of exposure. integration times and the range of wavelengths chosen yielded an average signal-to-noise ratio of nearly 800 in the ΔA absorbance spectrum. One approach is to limit the objective to determination of applied dose at the end of a beam segment. This approach overcomes the inter-exposure time dependence of OD on various parameters discussed above by measuring the OD at the end of exposure. According to the model in Figure 3, this event is marked by an abrupt change in OD increase, and can be assumed to occur when the δod/δt is half of the average (δod/δt) irrad (Figure 9). The average of the first five measurements thereafter can be taken as net change in OD for a given dose (i.e. derivative method) assuming OD is calculated with respect to the photons initially transmitted through unirradiated film, as specified above. An alternative, model method, was used. In this method, the data was fitted linearly using the least squares, yielding three lines. These represent OD(t)

51 s average 6 MV 381 cgy OD δod/δt 0.00 IRRADIATION Time (s) Figure 9. Change in optical density and rate of change in optical density for GafChromic MD- 55 film as a function of time before, during and after exposure to 381 cgy with 6 MV X-rays. prior to irradiation, during irradiation, and after irradiation. The intercept of second and third linear fit was found, and was taken as the time of termination of irradiation (Figure 10). With a justified assumption that the first fitted line is noise about OD=0, and the intercept between the first and second fitted lines is also zero within experimental error, the <OD> as given by the five points following termination of irradiation is what is reported as the change in OD (ΔOD) for a given radiation exposure. If the signal is not zeroed prior to each exposure (I 0s reset to the new amount of light intensity received, even though a dose might have been applied recently), and/or if the previously applied dose was large (~10 Gy), a slightly different approach can be taken. The ΔOD for the given dose can be easily measured by using the intercept of first and second lines of fit (Figure 11). That is, ΔOD for a given dose is then the <OD> measured immediately after intercept of second and third lines minus the <OD> measured immediately prior to intercept of first and second fit lines.

52 MV 381 cgy 0.21 m 2 OD m 1 Termination of Exposure y = m1x + b int = I y = m2x + b 1 2 Measured Fit m 1 Fit m Time (s) Figure 10. Termination of exposure is taken as the intercept of the two fitted lines: first line corresponding to data obtained during exposure and second line corresponding to data obtained after end of exposure ( denotes intercept). This approach would be valid for other media where OD versus dose is linear during exposure, and (δ(od)/ δt) irrad > (δ(od)/ δt) post-irrad, as seen for GafChromic MD-55 film. Finally, yet another approach is to use an independent detector to generate signal corresponding to radiation pulses in order to synchronize radiation delivery and radiochromic film data acquisition. This approach, however, would require a second dosimeter, adding to the complexity of the process, rather than simplifying it. As light-induced heating has the potential to induce local polymerization and contaminate the radiation-induced changes in OD,,(19) it was important to make sure the power delivered to the sampled spot on the film was sufficiently low to avoid errors introduced by this process. The approximate power delivered to the film in the phantom was estimated using a lab-grade system (Newport 840-C power meter, Newport 818-SL detector, an ORIEL Instruments Radiometric Dr. David Lewis, ISP ( ) (personal communication).

53 int( x, y ΔOD Dose =<OD> f -<OD> i f f y = m1 x + b1 ) = I y = m2x + b2 OD y = m 0 x + b0 int( xi, yi ) = I y = m1x + b MV 1 Layer of Film Time (s) Figure 11. Change in optical density can be calculated for any exposure by subtracting initial OD from final OD. Power Supply, ORIEL 250 W Quartz-Tungsten-Halogen (QTH) lamp in ORIEL Housing 66881, optical bandpass filter with 678 ± 10 nm FWHM were used) that replicated the bandpass of the LED described above but permitted greater accuracy in determining light transfer through the optical fibre, as the light coming out of delivery fibre was otherwise too dim to measure with other sources of light present in the room. The loss in power between the filtered QTH lamp and that delivered to film was then applied to the measured output of the LED operating at 30 ma. The LED, operating at 30 ma, outputs approximately 43 μw of light power resulting in approximately 7 nw at the end of the delivery fiber. The large power losses are mainly due to poor fiber-fiber coupling (simple SMA-to-SMA connectors are used), and inefficiency of coupling light out of 600-μm fiber into a 50-μm fiber. However, since efficiency of the optical system is not the focus of this study, no improvements have been implemented. The effect of this power on the film itself was measured using the above equipment and setup for a film with a pre-existing OD corresponding to an exposure of approximately 4 Gy.

54 38 The amount of light passing through phantom with a piece of GafChromic MD-55 film was recorded every 30 s over a period of one hour. The light intensity at the reference channel was collected simultaneously, and the procedure was repeated 6 times. The slope of mean OD in nm range as a function of time was calculated for both signal and reference channel in the six trials. The results for signal and reference <OD/s> were found to be (-3 ± 5) 10-6 and (- 2 ± 2) 10-6, respectively. Both are null results within error, and no significant difference between the two slopes was found (Type I error of 5%, where Type I error would occur if a hypothesis that the results are the same is rejected, even though it was true). (35) For exposures, the center of the GafChromic MD-55 film inside the phantom was located at the isocenter (100 cm from the source) of the linear accelerator (Varian 2100 EX), with the film plane parallel to the beam axis. A 10 cm 10 cm field of 6 MV X-rays was used. The doses and dose rates reported here are those to a small volume of water in the centre of the phantom. No corrections were applied to accommodate any perturbing effects associated with the small air-filled light transport cavities in the phantom and the fused silica optical fibers used (Figure 7). The linear accelerator used for the experiments was calibrated using recommendations from TG21, (36) and its performance was monitored using an independent ion chamber (37) over the same period (60 days) during which the experiments were performed. The standard deviation in linear accelerator output over this time was found to be approximately 0.2%. A. ΔOD of GafChromic MD-55 at Various Doses Small filmlets (1 cm 1 cm) cut from a single sheet of GafChromic MD-55 film were inserted in the cylindrical phantom and exposed to single doses of 0, 24, 48, 95, 190, 286 or 381 cgy at 286 cgy/min, with 3-6 films repeated at each dose level. All spectra acquisitions were initiated 5-10 seconds prior to exposure. The acquisitions for films exposed to 381 cgy were collected over a one-hour period. The spectra acquisitions for other films were collected over a period of 1-10 min. The ΔOD for each irradiation was calculated using the model method described above. The results of these measurements formed the dose versus ΔOD calibration plot. Previously unexposed film from either the same sheet, or another sheet within the same batch, were then each irradiated to known doses in the range of cgy (0, 67, 133, 200, 267, 333 cgy), with several individual films used per dose to evaluate precision. The calibration plot

55 39 was applied to the measured ΔOD to predict the applied radiation dose. All inferred doses for a given dose were averaged, and the values were compared to the applied doses. The alternative method in which a pre-exposure calibration of known dose was applied for each filmlet, instead of batch calibration, was also investigated. Each filmlet was first exposed to 95 cgy at 286 cgy/min, and sensitivity factor in cgy per ΔOD unit was recorded. Eight 48 cgy exposures were then applied at 5 minute intervals using the same dose rate to the same filmlet. Similarly, another piece of film was pre-exposed to 95 cgy, and then exposed to 190 cgy eight times. Changes in OD for each irradiation was calculated using the model method, and the inferred dose estimated using both the batch sensitivity factor and the individual pre-exposure sensitivity factor. B. Sensitivity as a Function of Layer Thickness It has been previously shown that increasing the number of layers of GafChromic MD- 55 increases the overall sensitivity. (12) The same should hold true for the system presented here. Two more sets of filmlet holders were created with capacity to layer two or four pieces of film within the apparatus. Experiments were performed with one, two, and four stacked pieces of previously unexposed pieces of film with continuous exposure at a dose rate of 286 cgy/min to a maximum OD of 2.5. This was repeated three times. To estimate the sensitivity of the system, the OD measured for the first ~1100 cgy in 1 and 2-layer setup was fit using the least mean squares. For the 4-layer setup, the data for the first 700 cgy was considered. The slope of the linear regression line is taken as sensitivity of the system. The sensitivity for each of the 1, 2 and 4-layer setups were compared. C. ΔOD of GafChromic MD-55 at Various Dose Rates The influence of dose rate was explored over a range from 95 to 571 cgy/min, by varying the pulse repetition rate and keeping dose per pulse the same. Filmlets were exposed to a specified dose at three rates: at 95 cgy/min, 286 cgy/min and at 571 cgy/min with approximately five minutes between irradiations. This was performed six times for the same dose, each time with a different permutation of dose-rate and dose (of 24, 48, 95, 190, 286 or 381 cgy). The mean ΔOD and mean δod/δt were calculated for each dose at the three different dose-rates, and the resulting values compared.

56 D. ΔOD Dependency on Temperature 40 To investigate the effect of temperature on the ΔOD, the phantom was placed into a waterproof cover (Endocavity Latex Ultrasound Transducer Cover, B-K Medical Systems Inc.) and submerged in a 1 L Pyrex beaker filled with ice water (Figure 12). The beaker was positioned on a heating plate (Corning Glass Works PC 351). The centre of the phantom was placed at isocentre and irradiated laterally by positioning the linear accelerator at 90. The temperature reported is that of the water bath, which was controlled using the heating plate and monitored with a thermometer. Once the desired temperature was reached, the phantom and film within were given approximately 15 minutes to equilibrate, which is assumed to be adequate for such a small phantom. The spectra were collected during and after exposure, and ΔOD reported versus temperature are those immediately at the end of exposure. Collection Fiber Delivery Fiber Latex Cover H 2 O Heating Plate Solid Water Phantom Figure 12. Schematic of setup for temperature dependency experiments. E. Continuous Versus Pulsed Irradiation The influence of pulsed and continuous radiation on the model method was examined. With a single filmlet within the phantom, the device was positioned at 81.5 cm surface-to-axis

57 41 distance (SAD) under a Cobalt-60 treatment unit. Using a 10 cm 10 cm field, and a dose rate of 85 cgy/min, the film was exposed six times for a period of one minute each, with approximately five minutes between exposures. A second single filmlet from the same sheet of film was exposed six times at an average dose rate of 85 cgy/min for a period of one minute each on the linear accelerator. Using the model method, both the intra-exposure slope and ΔOD were calculated. The values were then compared between the two methods of irradiation. It was assumed that the difference in instantaneous dose rate between the two irradiations yields a negligible difference in film response. III. Results A. ΔOD of GafChromic MD-55 at Various Doses The ΔOD of film versus time for a dose of 381 cgy is shown in Figure 13. The interexposure development appears to be logarithmic, while the OD growth during exposure seems linear (see inset)- note the abrupt change in rates of OD at the onset and completion of MV 381 cgy OD 0.10 OD IRRADIATION Time (s) Time (s) Figure 13. Change in optical density for GafChromic MD-55 exposed to 381 cgy with 6 MV X-rays as a function of time.

58 42 OD <OD> MV 381 cgy Time (s) 6 MV 381 cgy Time (s) Figure 14. Change in OD for five pieces of film, each exposed to 381 cgy with 6 MV X-rays (at the doserate of 285 cgy/min). The variability observed in OD at the end of dose delivery is due to non-uniformity of film response. The bottom plot illustrates the average of such measurements, with the error bar equivalent to σ (one standard deviation) MV Linear Fit (R = ) D i ( 1.03 ± 0.01) D + (1.5 ± 0.1) = g Inferred Dose (cgy) <%Error> Applied Dose (cgy) 0 Figure 15. Inferred dose using ΔOD measurements and calibration plot as a function of applied dose. The average percent error of inferred dose with respect to the applied dose is less than 5%. The error bars for both inferred dose and percent error data are equivalent to σ (one standard deviation).

59 MV 1 Layer of Film 0.8 OD Exposure Post/Pre-Exposure Time (s) Figure 16. Change in optical density for a piece of GafChromic MD-55 film during several exposures applied approximately 5 minutes apart. exposure. Figure 14 shows the OD(t) for five pieces of film, each exposed to 381 cgy with 6 MV X-rays. As the irradiation time and delivered dose increase, the deviation of OD increases as well. Using the average ΔOD for each given dose in 0-4 Gy range, the calibration line was calculated to be D(cGy) = (1 ± 2) + (1863 ± 22) ΔOD This calibration equation and ΔOD of a new set of freshly irradiated films (applied doses of cgy) were used to calculate inferred dose, shown versus applied dose in Figure 15. The linear fit yielded an intercept of 1.5 ± 0.1 cgy and a slope of 1.03 ± 0.01 (95% confidence, r>0.9999). The % error of dose prediction remains approximately constant with delivered dose. This seems counterintuitive at first, since one would expect the error to decrease as dose increases because of reduced percent error associated with performing a ΔOD measurement, as long as the number of photons transmitted through the film does not become too low. However, the largest contributing factor to the overall error in dose estimate comes from the variation in

60 44 film sensitivity, reported to be <4% (one standard deviation). * This variation contributes equally to low dose and high dose measurements. The other source of error is the systematic errors associated with using the calibration curve, which may be under- or overestimating the doses if the relationship deviates from linearity. Thus the decrease in percent error anticipated for higher doses may be compensated for by the systematic errors from deviation from signal linearity, making errors look approximately equal across the range tested. Figure 16 shows OD versus time for a single piece of film, which was exposed multiple times. The sharp increases in OD correspond to periods of intra-exposure to random doses, and the flatter regions are the data for inter-exposure. Using this ability of film to be exposed more than once and the pre-exposure calibration technique, inferred doses for 48 and 190 cgy irradiations are calculated. Table 3 compares the two methods of calculating inferred dose. The difference in error may appear striking at first, with the <%Error> being more than ten fold lower for 0.5 Gy than for 1.9 Gy. These errors are composed of the error from difference in film sensitivity (corrected in part by pre-calibration), and error in readout of ΔOD. A single film repetitively exposed to a fixed dose showed reproducibility of ΔOD of ~3.7%. The ΔOD/Gy decreased with increasing total dose delivered, deviating from the pre-exposure calibration factor that was established for the first fraction, and yielding an under-estimate in dose. Thus the average percent error for 1.9 Gy is higher than that for 0.5 Gy, which was only delivered a total dose of ~5 Gy by the end of this experiment. Table 3. Comparison of inferred dose and percent error using calibration plot and pre-exposure calibration as methods of calculation. Dose Calibration Plot Pre-Exposure Calibration (cgy) <Dose i (cgy)> <%Error> <% Error > <Dose i (cgy)> <%Error> <% Error > ± ± 2 47 ± ± ± ± ± ± 4 B. Sensitivity as a Function of Layer Thickness Figure 17 illustrates OD as a function of dose for one, two and four layers of stacked filmlets, with the average sensitivity of (5.4 ± 0.2) 10-4, (10.9 ± 0.4) 10-4, and (21.9 ± 0.2) 10-4 OD/cGy respectively. These correspond to an increase in sensitivity of 2.0 ± 0.1, 2.01 ± *

61 0.08 and 4.1 ± 0.2 for G 2-layer :G 1-layer, G 4-layer :G 2-layer and G 4-layer :G 1-layer respectively, which are all within error of anticipated increases of 2 and 4 times OD layer 2 layer 4 layer = (5.4 ± 0.2) Film Layer 2 Film Layers 4 Film Layers Dose (cgy) G G G 4 = (10.9 ± 0.4) 10 = (21.9 ± 0.2) 10 Figure 17. Optical density as a function of dose for a system utilizing one, two and four pieces of stacked film. G indicates signal gain, or the slope of the linear fit shown above. 4 4 C. ΔOD of GafChromic MD-55 at Various Dose Rates The average ΔOD measurements for a given dose at three different dose-rates are shown in Figure 18. Analysis of variance was performed on all measured ΔOD for a given dose. Since systematic errors may render statistical tests invalid, care was taken to eliminate any systematic errors that could be identified (rather than correct for them). The spectrometer linearity was checked using neutral density filters (Melles Griot, Carlsbad, CA) and the stability of the LED was also monitored during the experiments using a reference channel on the spectrometer to avoid collecting data if a drift in LED output was detected. Differences in background OD of the film and any variations in film setup within the phantom were eliminated by always using the light transmitted through unirradiated piece of film as initial intensity for ΔA and OD

62 46 calculations. To eliminate any trend or systematic error from film non-uniformity or LINAC output, the pieces of film were cut into squares prior to the beginning of experiment and taken out randomly from the package; the irradiations were then randomized with respect to the delivered dose and dose rate used. Thus, the random errors associated with the ΔOD measurements include variations from LINAC output, film response non-uniformity, deviations in dose from setup errors, and spectrometer noise. For 190 cgy, a statistically significant difference in ΔOD measurements was found (Type I error of 5%); however, no significant difference was found between the measurements obtained at different dose-rates for Type I error of 1%. (35) Nonetheless, a trend can already be seen in the ΔOD versus dose slopes calculated at each dose-rate, where the deviations quoted are those for 95% confidence interval. The slopes of OD/s as given by the model method calculation were also investigated (Figure 19). They were (8.7 ± 0.3) 10-4 s -1, (2.50 ± 0.01) 10-3 s -1, and (4.8 ± 0.2) 10-3 s -1 for 95, 285 and 570 cgy/min respectively. The slope ratios are 2.8 ± 0.2, 2.0 ± 0.2 and 5.6 ± 0.5 for D 285 :D 95, D 570 :D 285 and D 570 :D 95 respectively. For a given dose-rate, the slopes ΔOD (5.6 ± 0.1) D ΔOD = (5.4 ± 0.1) D (5.3 ± 0.1) D R> cGy /min 286cGy / min 571cGy /min + ( 10 ± 7) + ( 8 ± 5) 10 + ( 5 ± 7) 4 6 MV ΔOD at 95cGy/min ΔOD at 286cGy/min ΔOD at 571cGy/min Dose (cgy) Figure 18. Change in optical density as a function of dose for doses delivered at 95 cgy/min, 286 cgy/min and 671 cgy/min.

63 OD/s (x10-3 ) MV 95 cgy/min 286 cgy/min 571 cgy/min Dose (cgy) 47 Figure 19. Rate of change in optical density as given by the linear fit of data obtained during exposure as a function of applied dose, for doses delivered at 95 cgy/min, 286 cgy/min and 571 cgy/min. are within error of each other for doses in the cgy range, even though the trend is for a higher slope estimate to appear at higher doses. An analysis of variance showed that there was a significant difference (Type I error of 5%) between OD/s estimates for different doses delivered at 95 cgy/min and 571 cgy/min, but not at 286 cgy/min. At Type I error of 1%, only OD/s estimates for 95 cgy/min exhibited a significant difference between given doses. (35) D. ΔOD Dependency on Temperature Using the nm averaging window, ΔOD for a consistent dose was measured at each temperature. No effort was put into calculating the actual dose for this setup, but the exposures were all done at 300 Monitor Units per minute for one minute. The mean ΔOD increases from ± at 0.0 ± 0.5 C, to ± at 31.5 ± 0.5 C, and then

64 MV 680 λ max (nm) C 20 C 38 C 660 ΔA Wavelength (nm) Temperature ( C) Figure 20. Position of wavelength of maximum absorbance for GafChromic MD-55 as a function of irradiation/measurement temperature. ΔOD MV ~10 nm about λ max nm window Temperature ( C) Figure 21. Change in OD for a given dose as a function of applied/measured temperature using both a constant spectral averaging window and a shifting spectral averaging window.

65 49 decreased to ± at 43.0 ± 0.5 C. Figure 20 shows a decrease in λ max to lower values with increasing temperature. To eliminate the temperature effect as purely that due to shifting of wavelength of maximum absorbance (λ max ), the average λ max (<λ max >) was found for each irradiation at a given temperature, and the ΔOD was recalculated with the 10 nm averaging window approximately about this <λ max > (Figure 21). Averaging of OD about λ max instead of a fixed window increases measured ΔOD for a given dose when λ max was on the periphery or outside of the static averaging window. E. Continuous Versus Pulsed Irradiation The mean regression slope (of the linear fit performed during application of model method) and ΔOD obtained for a film irradiated to 85 cgy at 85 cgy/min with Cobalt-60 are (7.34 ± 0.12) 10-4 s -1 and ± respectively. The mean regression slope and ΔOD obtained for a film irradiated to 85 cgy at 85 cgy/min with the linear accelerator are (7.47 ± 0.15) 10-4 s -1 and ± respectively. The ratio of the average slope obtained on a linear accelerator to average slope obtained on Cobalt-60 unit is 1.02 ± The ratio of the average ΔOD obtained on a linear accelerator to average ΔOD obtained on the Cobalt-60 unit is 1.00 ± Both ratios are within error of Hence no difference in slope and ΔOD values between the two modes of dose delivery was found for a given dose delivered at the same dose rate. IV. Discussion The utilization of MD-55 for real-time, point-based dosimetry appear feasible based on the results presented here. There are a number of points that need to be raised with respect to these results: the relative stability of the sensitivity of the system (film in combination with optical path) is one of the significant sources of errors; the dynamic range for multiple-layer systems can be increased by modifying the light source and decreasing spectrophotometer noise; interexposure development introduces small errors to measurements for doses of ~2 Gy and above; temperature fluctuations in the range of ºC may be responsible for variation in ΔOD measurements of nearly 10%.

66 A. OD of GafChromic MD-55 at Various Doses 50 The sensitometric response difference <8% from the mean within a single sheet, and <5% from the mean between sheets of a single batch is expected. The deviation of the ΔOD values seen for a single delivered dose can be easily explained by the sensitometric response variation present in this product due to uneven thickness of layer, and manufacturing process. While the fits performed in the calibration process are encouraging, there are some subtle elements. The fit of the inferred dose versus applied dose showed an intercept of 1.5 ± 0.1 and a slope of 1.03 ± 0.01 (95% confidence). Neither the intercept nor the slope is within error of the desired values of 0 and 1.00 respectively. One possible source of the error on the intercept is one introduced by the calculation procedure, where a non-zero dose is assumed to have been delivered, because the algorithm is not set up to distinguish true signal from noise. So even when a dose of zero Gy was delivered, a non-zero dose was still assigned. A more accurate procedure can be conceived which would include a minimum OD threshold value, below which any signal is assumed to be merely noise, and dose will not be calculated. This threshold value would be based on the average of signal fluctuation expected for a given setup. A closer look at percent error of each inferred dose with respect to the given dose shows a consistent overestimate, resulting in an average percent error of 3-5% for each dose in question. The percent error is often within the 8% sensitometric variation of a single sheet of film. Although the 1cm 1cm pieces were picked randomly once cut, one half of the 5 5 film was cut first for the calibration, and the second half was cut later for the testing of the method. This would suggest that the variation in sensitometric response of GafChromic MD-55 film is not random throughout the sheet, but may be observed to increase in a single direction, likely a byproduct of the manufacturing process. (38) Table 3 shows the comparison between using a calibration plot and pre-exposure calibration to infer a dose. With the former, the inferred dose for 48 cgy tends to be overestimated on average, producing the mean percent error of 3 ± 2%. While using the preexposure calibration produces the same mean percent error, on average the inferred doses are underestimated. The standard deviation of a measured OD (average of 5 data points at the end of exposure) for a 48 cgy dose ranged between units. Such low deviations in OD illustrate that the amount of light getting to the spectrophotometer is sufficient to keep the noise reasonably low and close to the best possible precision of (using this 12-bit Analog-to- International Specialty Products (ISP) product information.

67 51 Digital Converter). According to the calibration plot, this standard deviation of corresponds to 1-2 cgy, or approximately 2-4% for 48 cgy dose. Hence, it is not realistic that given such fluctuations in OD for a single irradiation, we should expect errors lower than that. The deviations on OD measurements for 190 cgy irradiations were units, corresponding to a dose error of 2-5 cgy, or 1-3 % error. The observed errors are larger than that, with both methods underestimating the dose. The remainder of the error comes from two sources: loss of sensitivity and decreased polymerization kinetics as the film accumulates dose. Thus as the number of dose fractions is increased, the ΔOD for the given dose decreases, adding systematic error to the measurement. This effect is more pronounced for the higher dose fractions, as the total dose given to the film is greater than for low dose fractions. The pre-exposure calibration technique had a lower error on dose estimate, decreased the mean percent error of the estimates by a factor of two, correcting for the variation in film sensitivity and to polarization effect of the film itself. (16) The pre-exposure technique described would also be useful to account for change in sensitivity as the dosimeter ages, since OD increases with storage time, (16) and response of film with dose depends on the total dose applied or OD to date. (13) For a clinical dosimeter using a sensitive medium such as GafChromic MD- 55, this pre-exposure calibration would extend storage time after which the dosimeter can still be accurately used. B. Sensitivity as a Function of Layer Thickness It was shown that increase in sensitivity occurred with increased thickness of ionizingradiation sensitive layer. However, unless the light intensity, integration time or number of averages taken is increased, the signal reaches the saturation level (spectrometer does not detect enough light to distinguish further increase in optical density, and OD signal levels off) quicker for increased number of layers. A positive side effect of increasing layer thickness is also seen in reduced deviation of OD measurements from linearity over the usable OD range (before the optical system reaches saturation of OD>2.5). It is known that GafChromic MD-55 film has a range of linear response for ΔOD versus dose, which depends on wavelength of measurement. (7-11) Near the peak of maximum absorbance (~675 nm), this range is the shortest, partly due to shift of λ max with dose, and also due to loss of sensitivity as the polymer-to-monomer ratio within the crystal increases. This effect is seen for the OD versus dose curve for a single layer of film, but not seen for four layers of film. The simple explanation is that by the time saturation for this optical system is reached (OD>2.5), the dose absorbed by each of the sensitive layers in the four-

68 52 film system is approximately 4 times less than the dose absorbed by a single-film system at the same OD value. Hence, the response of the film in four-film system to ionizing radiation dose does not deviate from linearity over the range of OD used. It is possible to increase the dynamic range of a multiple-layer system by using a lower noise, higher resolution spectrophotometer with a larger absorbance range, and by increasing the amount of incident light in the appropriate wavelength range. Implementation of a new light source must be done with caution, especially if continuous measurements are performed, as done for the experiments described here. Increasing the power of light source may induce changes in OD due to local heat polymerization from the light source itself. C. ΔOD of GafChromic MD-55 at Various Dose Rates Each fraction of ionizing radiation causes a rapid change in OD of GafChromic MD-55 medium while local concentration of polymer is low, followed by a slow increase in OD due to increasing separation between adjacent monomers as polymerization proceeds and local concentration of polymers increases. If one waits hours after irradiation as recommended, then most of inter-exposure development will have completed. However, if the measurements are done immediately after the end of irradiation, the amount of inter-exposure development that has completed for a given dose depends on the time it took to give that dose, or the dose-rate. Our results show that ΔOD measurements for a given dose at three different dose-rates of 95 cgy/min, 286 cgy/min and 571 cgy/min are indistinguishable (Type I error of 1%) for all doses up to 381 cgy. Doses 95 cgy showed no dose-rate dependence (Type I error of 5%), but those of 190 cgy either exhibited dose-rate dependence or were close to the critical F value. The F value is the average squared difference between means of sets, divided by the sampling variation expected; if the calculated F value is greater than the critical or expected F value for a total given number of data points, number of sets and type I error, then there is a statistically significant difference between the sets of data. (35) With our experimental setup and errors in ΔOD measurements, differences between sets of measurements done at different dose-rates can be largely explained by fluctuations within the measured values themselves. At higher doses, the variation in ΔOD for various dose-rates can introduce a substantial error. The ratios of average OD/s estimates are within error of the respective dose-rates, showing that the rate of OD increase during exposure is in fact proportional to the applied doserate. The slopes (OD/s), as given by the linear fit of each curve during data processing, on average showed a higher slope estimate at higher doses. This is consistent with the presence of a

69 53 slow-kinetics increase in OD towards the end of a single polymer chain. For a given dose-rate, the higher dose will have a longer irradiation period, and hence a greater inter-exposure development due to the first few fractions of the overall given dose. This extra OD will result in a higher slope (OD/s) when linear fit is performed during calculation of OD for this dose. To investigate how important the inter-exposure development from the first few fractions to the performance of the fitting algorithm is, an analysis of variance on OD/s values for each dose-rate was performed. Results showed that there was a significant difference (Type I error 5%) between OD/s estimates for different doses delivered at 95 cgy/min and 571 cgy/min, but not at 286 cgy/min. (35) The discrepancy in OD/s estimates for the lowest dose-rate can be easily explained. The time difference between delivering 286 cgy and 381 cgy is a full minute, and hence the larger the dose given, the greater OD due to inter-exposure development has occurred. The linear regression performed during model method calculation will then take these higher values of OD into account, giving a larger OD/s estimate. The OD/s estimates for highest doserate may be out of error with each other due to fitting algorithm: the number of data points used in the fit at the highest dose-rate is less (2 or 6 times less) than the number of points used at the other dose-rates for the same given dose. It is quite likely that changing the dose per pulse will have an effect on polymerization kinetics, and thus changing dose rate by varying dose per pulse (instantaneous dose rate) instead of pulse repetition (average dose rate) will yield different results than described above. A study by McLaughlin et al showed termination of propagation of polymerization within 2 ms of irradiation. (39) The experiment was performed by irradiating films to 20 Gy with a 50 ns pulse, using an instantaneous dose rate significantly greater than that used clinically. When clinical instantaneous dose rates are used, propagation of polymerization can last hours and even days after irradiation. Dependence of polymerization kinetics on instantaneous dose rate is outside the scope of this paper, but should be investigated in the future. More specifically, an instantaneous dose rate value at which polymerization appears to occur significantly faster, leading to almost immediate optical saturation or complete polymerization of all monomers within the system, should be established. D. ΔOD Dependency on Temperature Our results, which show an overall decrease in ΔOD for a given dose as one increases from approximately room temperature to over 40 C, is consistent with previously published data. Part of the sharp decrease, once temperature reaches over 32 C, and the large fluctuations within

70 54 the temperature range are due to the position of λ max in the spectrum, and the choice of wavelength of interrogating light or spectral averaging window. The wavelength of maximum absorbance was found to decrease with temperature, and the λ max observed at 20 and 32 C are similar to those reported earlier by Klassen. (16) Averaging of OD about λ max instead of a fixed window increases the measured ΔOD for a given dose if λ max is on the periphery or outside of the static averaging window. However, even with λ max shift taken into account, an increase in ΔOD with temperature is seen between 0 and 30 C, followed by a decrease in ΔOD. The latter can be explained by loss of monomer crystal structure that occurs as temperatures close to 60 C are approached. The imperfect alignment of the monomers may alter unit cell parameters within the crystal and decrease the chance of polymerization. (40) It is clear that changes in temperature of the dosimeter introduce new errors that are not accounted by using either a calibration plot or a pre-exposure calibration done at a set range of wavelengths. Some of the errors can be eliminated if the averaging range for OD measurements are done over 10 nm about the temperature dependent λ max. The tracking of λ max might also reduce some of the error of using GafChromic MD-55 dosimeter after a long shelf life or after several large exposures, since λ max varies with storage time and total dose given. (2,7,14,15,17) While OD measurements can be made nearly independent of read-out temperature by using nm, (18) changing the temperature during exposure and real-time readout introduces variations in chemical reaction rates that may ultimately lead to irreversible effects in OD. The source of the remaining error, which can still be as high as 8%, is possibly due to loss of crystal structure upon temperature increase, which can affect sensitivity to ionizing radiation. (20,21) The aim for the dosimeter is to have broad applications, and often the temperature at which the dosimeter will be used cannot be known in advance. Requiring prior knowledge of the temperature in order to perform a pre-exposure calibration at that exact temperature significantly complicates its use and increases cost. A simpler dosimeter needs to have an insignificant temperature dependence over 20-38ºC range so that it can be calibrated at room temperature and require no use of correction factors. E. Applications Although the apparatus and methods described above work sufficiently well for testing suitability of radiochromic media for real-time dosimetry, the system would not qualify as an in vivo dosimeter with its current design. An example of a more suitable in vivo prototype would

71 55 use the radiochromic suspension on a tip of an optical fibre, sandwiched between the fibre and a reflective medium, such as reflective paint or prism. Two fibres, one delivering light to the radiochromic medium and one collecting the transmitted light, can also be used instead of a single fibre approach. The entire system comprising of optical fibre(s), radiochromic medium, and reflector, would be encompassed in a biocompatible sheath to enable insertion into the tissue. The probe may be inserted via a small needle, using ultrasound guidance for positioning. The desired real-time in vivo dosimeter would meet all of the criteria discussed in Table 2 and would provide various improvements on existing dosimetry systems. One of the benefits of real-time dosimetry is the immediate access to dose information, unavailable with systems that require processing (e.g. thermo-luminescent dosimeters). The other significant improvement of the intended in vivo dosimeter is its near water-equivalency, which would eliminate the need for separate calibrations for energies that may be in use. This would allow the dosimeter to be used for various procedures, including verification of dose in areas that have unpredictable x-ray spectra (e.g. near small bony structures and air cavities that are relevant in some head and neck cancers), in the penumbra regions of large fields (where it would not overestimate dose due to low-energy over-response common in other dosimetry methods), (8,41) in brachytherapy treatments with a variety of sources of different energies, during IGRT, fluoroscopy and possibly others, while referring to a single calibration performed at a convenient energy of choice (such as Cobalt-60). V. Conclusion The change in optical density of GafChromic MD-55 as a function of time shows a rapid increase during exposure, with a rate that is proportional to applied dose-rate. The end of exposure is marked by an abrupt decrease in rate of OD, with the OD increasing thereafter slowly with time, approaching a value that would typically be measured hours later and related to dose. It was found that ΔOD measured immediately at the end of exposure is proportional to dose in the tested range of 0 to 4 Gy. Performing such measurements can predict the applied dose immediately at the end of exposure with an average error of 5% for Gy doses. Calibrating individually each dosimeter prior to its use can decrease the errors. To increase precision of low-dose predictions, the sensitivity of the dosimeter can be proportionally increased by increasing the thickness of sensitive layer. For dose rates varying between 95 cgy/min and 570 cgy/min, it was found that measurements of ΔOD immediately after the end of exposure did not depend on dose-rate for low doses ( 100 cgy), typical of those delivered in a

72 56 single beam during fractionated external beam radiation therapy. Higher doses ( 2 Gy) exhibited some dose-rate dependence. While GafChromic MD-55 appears to be suitable for real-time measurements of ionizing radiation dose, numerous issues remain. The ΔOD measurements performed immediately at the end of exposure exhibited significant temperature dependence in the clinically relevant range of C, which could not be corrected by shifting the spectral window of interest with the decreasing wavelength of maximum absorbance. Overall, this approach, which employs existing GafChromic MD-55, shows the promise of the novel read-out configuration and method and has revealed some subtleties of the sensitive material in GafChromic MD-55 that were previously unreported. ACKNOWLEDGEMENTS The authors wish to thank the following for their contribution: Yuen Wong and Brian Taylor, Robert Rothwell, Robert Rusnov, Dr. Robert Heaton, Cameron Chiarot, Dr. Robert Weersink and Photonics Research Ontario, and Dr. Douglas Moseley. This work was in part funded by National Institute of Health / National Institute on Aging (R21/R33 AG19381) and by the Fidani Center for Radiation Physics. 1. Miller A., McLaughlin W.L. Absorbed dose distribution in a pulse radiolysis optical cell. International Journal for Radiation Physics and Chemistry 7: 661-6, McLaughlin W.L., Yun-Dong Chen, Soares C.G., Miller A., Van Dyk G., Lewis D.F. Sensitometry of the response of a new radiochromic film dosimeter to gamma radiation and electron beams. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 302: , McLaughlin W.L., Puhl J.M., Al-Sheikhly M., Christou C.A., Miller A., Kovacs A., Wojnarovits L., Lewis D.F. Novel radiochromic films for clinical dosimetry. Radiation Protection and Dosimetry 66: 263-8, McLaughlin W.L., Al-Sheikhly M., Lewis D.F., Kovacs A., Wojnarovits L. A radiochromic solid-state polymerization reaction. In: Irradiation of Polymers:

73 Fundamentals and Technological Applications. Washington: American Chemical Society, p , McLaughlin W.L., Soares C.G., Sayeg J.A., McCullough E.C., Kline R.W., Wu A., Maitz A.H. The use of radiochromic detector for the determiantion of stereotactic radiosurgery dose characteristics. Medical Physics 21: , Dutreix A., Bridier A. Dosimetry for External Beams of Photon and Electron Radiation. In: K.R.Kase, B.E.Bjarngard, F.H.Attix, eds. The Dosimetry of Ionizing Radiation. Toronto: Academic Press, Inc., Saylor M.C., Tamargo T.T., McLaughlin W.L. A thin film recording medium for use in food irradiation. Radiation Physics and Chemistry 31: , Muench P.J., Meigooni A.S., Nath R., McLaughlin W.L. Photon energy dependence of the sensitivity of radiochromic film and comparison with silver halide film and LiF TLDs used for brachytherapy dosimetry. Medical Physics 18: , Chiu-Tsao S., Zerda A., Lin J., Kim J.H. High-sensitivity GafChromic film dosimetry for 125 I seed. Medical Physics 21: 651-7, Reinstein L.E., Gluckman G.R. Comparison of dose response of radiochromic film measured with He-Ne laser, broadband, and filtered light densitometers. Medical Physics 24: , Reinstein L.E., Gluckman G.R. Predicting optical densitometer response as a function of light source characteristics for radiochromic film dosimetry. Medical Physics 24: , Cheung T., Butson M.J., Yu P.K.N. Use of multiple layers of GafChromic film to increase sensitivity. Physics in Medicine & Biology 46: N235-N240, Ali I., Costescu C., Vicic M., Dempsey J.F., Williamson J.F. Dependence of radiochromic film optical density post-exposure kinetics on dose and dose fractionation. Medical Physics 30: , 2003.

74 Chu R.D.H., Van Dyk G., Lewis D.F., O'Hara K.P.J., Buckland B.W., Dinelle F. GafChromic dosimetry media: A new high dose, thin film routine dosimeter and dose mapping tool. Radiation Physics and Chemistry 35: , Mack A., Mack G., Weltz D., Scheib S.G., Böttcher H.D., Seifert V. High precision film dosimetry with GafChromic films for quality assurance especially when using small fields. Medical Physics 30: , Klassen N.V., van der Zwan L., Cygler J. GafChromic MD-55: Investigated as a precision dosimeter. Medical Physics 24: , Meigooni A.S., Sanders M.F., Ibbott G.S., Szeglin S.R. Dosimetric characteristics of an improved radiochromic film. Medical Physics 23: , Reinstein L.E., Gluckman G.R. Optical density dependence on postirradiation temperature and time for MD-55-2 type radiochromic film. Medical Physics 26: , Sullivan P.R., Hasson B.F., Grossman C.H., Simpson L.D. Optical density changes of GafChromic MD-55 film resulting from laser light exposure at wavelengths of 671 nm and 633 nm. Medical Physics 27: , Baughman R.H. Solid-state synthesis of large polymer single crystals. Journal of Polymer Science: Polymer Physics Edition 12: , Baughman R.H., Yee K.C. Solid-state polymerization of linear and cyclic acetylenes. Journal of Polymer Science: Macromolecular Reviews 13: , Guillet J. Photopolymerization. In: Polymer Photophysics and Photochemistry: an introduction to the study of photoprocesses in marcomolecules. New York: Cambrdige University Press, p , Sixl H. Electronic Structures of Conjugated Polydiacetylene Oligomer Molecules. In: Kuzmany H., Mehring M., Roth S., eds. Electronic Properties of Polymers and Related Compounds. Berlin: Springer Verlag, p , 1985.

75 Bloor D., Koski L., Stevens G.C., Preston F.H., Ando D.J. Solid state polymerization of bis-(p-toluene sulphonate) of 2,4-hexadiyne-1,6-diol. Journal of Materials Science 10: , Sixl H., Warta R. Excitons and Polarons in Polyconjugated Diacetylene Molecules. In: Kuzmany H., Mehring M., Roth S., eds. Electronic Properties of Polymers and Related Compounds. Berlin: Springer Verlag, p , Tsibouklis J., Pearson C., Song Y.P., Warren J., Petty M., Yarwood J., Petty M.C., Feast W.J. Pentacosa-10,12-diynoic Acid/Henicosa-2,4-diynylamine alternate-layer Langmuir- Blodgett films: synthesis, polymerization and electrical properties. Journal of Materials Chemistry 3: , Solomons G., Fryhle C. Organic Chemistry. 7 ed. Toronto: John Wiley & Sons, Inc., O'Donnell J.H., Sangster D.F. Principles of Radiation Chemistry. London, England: Edward Arnold Ltd., Huston A.L., Justus B.L., Falkenstein P.L., Miller R.W., Ning H., Altemus R. Remote optical fiber dosimetry. Nuclear Instrumets and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms 184: 55-67, Becks K.H., Drees J., Goldmann K., Gregor I.M., Heintz M. A multichannel dosimeter based on scintillating fibers for medical applications. Nuclear Instrumets and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 454: , Bueker H., Haesing F.W. Fiber-optic radiation sensors. SPIE 2425: , Gripp S., Haesing F.W., Bueker H., Schmitt G. Clinical in vivo dosimetry using optical fibers. Radiation Oncology Investigations 6: 142-9, Day M.J., Pitchford W.G. The normalized peak scatter factor and normalized scatter functions for high energy photon beams. In: Aird E.G.A., Burns J.E., Day M.J., Duane S., Jordan T.J., Kacperek A., Klevenhagen S.C., Harrison R.M., Lillicrap S.C., McKenzie A.L., Pitchford W.G., Shaw J.E., Smith C.W., eds. British Journal of Radiology,

76 Supplement 25: Central Axis Depth Dose Data for Use in Radiotherapy. London: The British Institute of Radiology, p , Fusi F., Mercatelli L., Marconi G., Cuttone G., Romano G. Optical characterization of a radiochromic film by total reflectance and transmittance measurements. Medical Physics 31: , Milton J.S., Arnold J.C. Introduction to Probability and Statistics: principles and applications for engineering and the computing sciences. 2 ed. Toronto: McGraw-Hill, Inc., Schulz R.J., Almond P.R., Cunningham J.R., Holt J.G., Loevinger R., Suntharalingam N., Wright K.A., Nath R., Lempert G.D. A protocol for the determination of absorbed dose from high-energy photon and electron beams. Medical Physics 10: , Martell E., Galbraith D., Munro P., Rawlinson J.A., Taylor W.B. A flatness and calibration monitor for accelerator photon and electron beams. International Journal of Radiation Oncology Biology Physics 12: 271-5, Butson M.J., Yu P.K.N., Cheung T., Metcalfe P.E. Radiochromic film for medical radiation dosimetry. Materials Science and Engineering:Reports R41: , McLaughlin W.L., Al-Sheikhly M., Lewis D.F., Kovacs A., Wojnarovits L. A radiochromic solid-state polymerization reaction. American Chemical Society. Papers presented at the Washington, D.C.Meeting [35], The Division of Polymer Chemistry, Inc. 40. Baughman R.H. Solid-state reaction kinetics in single-phase polymerizations. Journal of Chemical Physics 68: , Wang B., Kim C., Xu X.G. Monte Carlo modeling of high-sensitivity MOSFET dosimeter for low- and medium-energy photon sources. Medical Physics 31: , 2004.

77 CHAPTER 3: REAL-TIME RESPONSE OF GAFCHROMIC EBT Portions of the following have been published as Characterization and Real-Time Optical Measurements of the Ionizing Radiation Dose Response for a New Radiochromic Medium by Alexandra Rink, I. Alex Vitkin, and David A. Jaffray in Medical Physics 32(8), p (2005) 61

78 62 I. Introduction Several radiochromic films, manufactured by International Specialty Products (ISP, Wayne, New Jersey, USA) for dosimetry purposes, have been studied for more than a decade, (1-3) often for 2D radiation dose verification, (4,5) and thorough reviews have been published. (6,7) In general, a dose of 1 Gy or more is recommended for accurate dose measurement, (8) preventing use of these films for low-dose studies. Radiochromic substances have also recently been investigated in real-time dosimetry systems. (7) In these studies, GafChromic MD-55 performed reasonably well while some issues remained unresolved. GafChromic EBT film (ISP) was created for use in external beam dose verification, and is advertised by the manufacturer to be more sensitive than both GafChromic HS and MD-55 films. This makes it potentially useful for low dose verification, such as doses delivered in each IMRT segment. It is also proposed as an improvement for some of the real-time dosimetry issues, including stability of wavelength of maximum absorbance (λ max ), and decreased extent of post-exposure darkening. The radiochromic material used in EBT film appears to be a reasonable candidate for use as an optical media for real-time in vivo dosimetry, as per criteria previously listed. (7) It addresses the requirement for a small radiation sensitive volume with improved waterequivalency compared to its predecessor MD-55, increased sensitivity to low doses, and faster polymerization kinetics, resulting in a stable response shortly after exposure (ISP product information). To determine the suitability of EBT to real-time in vivo dosimetry and verify some of the above claims, an experimental set-up and method previously used in real-time investigations of MD-55 film was employed. (7) In this technical report investigations of linearity, stability, and sensitivity are described. II. Method and Materials The experimental setup described previously (7) was modified by removing the beam splitter (Figure 22) and using a light emitting diode (HLMP-ED25-TW00, Agilent Technologies, Palo Alto, CA, USA) with a ~633 nm emission peak (Figure 23), chosen to interrogate the greater of the two absorbance peaks of the EBT film. The EBT film used in these investigations (Lot # W) consisted of five layers (Figure 24, ISP product information), where the clear polyester was assumed to be Mylar as employed in the MD-55 film. (2) The 17 μm thick radiation D. Lewis, ISP (private communications, 2004 and 2005).

79 63 Film in Solid Water TM Phantom Optical Fibers (~ 17 m) Radiation Barrier Light Source Light Detector Processor Figure 22. Schematic of experimental setup. Detector is an Ocean Optics Inc. SD2000 dualchannel spectrophotometer. Processor is a computer with Ocean Optics Inc. IOBase32 and Matlab 6.1. LED Output (arb. units) Wavelength (nm) Figure 23. Emission of the light emitting diode used in experimental setup, as measured by spectrometer.

80 64 97 μm clear polyester 17 μm 6 μm 17 μm radiation sensitive layer deposition layer radiation sensitive layer 97 μm clear polyester Figure 24. Schematic of layers in EBT film. The overall atomic composition of this configuration is: 42.3% C, 39.7% H, 16.2% O, 1.1% N, 0.3% Li, 0.3% Cl (ISP product information). sensitive layers contain a suspension of radiochromic structures within a gelatin layer. It is the absorbance within these two layers that gives rise to the overall darkening of the film. The system operated with ± 0.01 ma power supply driving light output from the light emitting diode. A dark spectrum (I D ) and a reference spectrum (I R ) were collected prior to each radiation exposure. The I R at a given wavelength is proportional to the radiant power of the light transmitted through the unirradiated piece of EBT film (measured over ~0.33mm 2 ). (7) Irradiation of the film was initiated shortly (typically < 3 s) after starting the collection of the sample spectra (I S ). The spectrometer integration time ranged from 8-10 ms, depending on the study performed, and each spectrum was recorded. The change in absorbance (ΔA), at any measurement interval, for each wavelength was calculated using I R ( λ) I D ( λ) ΔA ( λ) log 10 (1) I S ( λ) I D ( λ) The EBT film likely undergoes interactions with interrogating light other than absorption. However, the optical scattering was assumed to be negligible, and the fraction of light reflected from both the clear polyester layer and from the sensitive layer was assumed to be independent of dose. Hence, the change in absorbance as measured by light transmission is assumed to be entirely due to radiation induced increase in concentration of absorbers within the sensitive layers of the film.

81 65 The method is illustrated in Figure 25, which shows the change in absorbance of a single piece of EBT film prior to exposure, immediately after exposure to 2.38 Gy at 2.86 Gy/min, and at 15 and 60 minutes after the completion of exposure. The spectral window of interest, or range for optical density calculation, is a 10 nm band around the main peak ( nm). The optical density (OD) was then defined as OD 1 n 1 ΔA( λi ) + ΔA( λi+ 1) ( ) ( λ 1 ) i 1 i λi (2) = λ λ 2 + n 1 where λ 1 to λ n are wavelengths that span the window of interest in the spectrum, sampled 3 times per nm. The light power incident onto a ~650 μm diameter spot (7) of the film within the holder was measured (840-C power meter and 818-SL detector, Newport, Mountain View, CA, USA) to be ~75±10 nw. The effect of this power level of interrogation light on GafChromic EBT film was investigated by monitoring the OD for an unexposed piece of film over a period of one hour on eight separate occasions. The average increase in OD was ± 0.002, and considered statistically insignificant (type I error of 5%, where type I error would occur if a MV 2.38 Gy 2.86 Gy/min Prior to exposure Immediately after exposure 15min after exposure 1 hr after exposure 0.8 ΔA Spectral averaging window Wavelength (nm) Figure 25. Change in absorbance of EBT film over a range of wavelengths before, immediately after, and at two time points post-exposure.

82 66 hypothesis stating that the increase in OD is null is rejected, even though it was actually true). (9) As in the previous experiments, systematic errors are assumed to be negligible, as the response of spectrometer was shown to be linear, and stability of LED was monitored. Thus the fluctuations described above are believed to be due to spectrometer noise. The radiation exposures were performed using the same setup as described in real-time investigations of MD-55. (7) A. ΔOD of EBT Film Versus Time Five 1cm 1cm pieces of EBT film were exposed to 9.52 Gy at an average dose rate of 2.86 Gy/min. The transmitted spectra were collected during exposure, and for approximately one hour after completion of exposure. The ΔOD values were calculated as described above and plotted versus time to investigate suitability of obtained signal with respect to a Fast Kinetics model. (7) B. Sensitivity and Stability Comparison Between EBT and MD-55 Films One 1cm 1cm piece of EBT and four stacked 1cm 1cm pieces (to increase optical signal and minimize error due to small fluctuations in light signal) (7,10) of MD-55 (lot #L1906MD55) were each exposed to a dose of 1.9 Gy. The MD-55 film was optically interrogated with a 680 nm light emitting diode (Roithner Lasertechnik, Vienna, Austria), using the same set-up as illustrated in Figure 22, and the same spectral range as in previous investigations ( nm). (7) The 33.0 ± 0.3 nw of power delivered to a ~650 μm spot on the previously un-irradiated film was shown to have a small effect on OD ( over 24 hours), but was tolerated in order to keep the signal intensity and integration time on the spectrophotometer sufficient for real-time measurements. For both types of radiochromic film, spectra were obtained during exposure and for approximately 19 hours following exposure without disturbing the system. The data obtained for MD-55 film was divided by four (corresponding to the factor of 4 increase in sensitivity anticipated for the 4 layers used in this study) in order to obtain the average OD increase for each individual film. To compare stability between the two types of films, the post-exposure measurements were normalized by the change in optical density measured immediately at the end of exposure.

83 67 C. Dependence of Real-Time OD Measurements on Dose Rate for the EBT Film Dose rate dependence of OD measurements performed during, or immediately at the end of, radiation exposure was investigated by irradiating EBT film to the same dose of 9.52 Gy at one of two different dose-rates (5.71 Gy/min or 0.95 Gy/min). Five films in total were exposed at each dose rate. The OD as a function of time was converted to a function of dose during exposure, using average dose rates as listed above. The values recorded immediately at the end of exposure were compared (analysis of variance using type-i error α = 0.01). (9) D. Structure of Active Crystals in MD-55 and EBT Films The active layer suspensions from MD-55 and EBT films (provided by Dr. Lewis, ISP) were imaged at the Advanced Optical Microscopy Facility, (Ontario Cancer Institute, Toronto, Canada). A differential interference contrast and Plan-Apo 63x/1.4 NA lens on an inverted microscope (Axiovert 200M, Carl Zeiss, Oberkochen, Germany) were used, providing 0.22 μm resolution. III. Results and Discussion A. OD of EBT Film Versus Time If radiochromic substances are to be used in real-time dosimetry, errors due to postexposure darkening have to be accounted for. The simplest way to eliminate a substantial fraction of post-exposure darkening is to limit measurements to those performed during exposure. This requires knowing when the radiation is present. Figure 26 shows OD versus time during 200 s exposure to 9.52 Gy (at 2.86 Gy/min) and for approximately one hour after exposure for a single piece of EBT film. A distinct difference in darkening rates between exposure and post-exposure (intra- and inter-exposure, respectively) can be seen, and duration of radiation exposure can be easily deduced from the signal. Change in optical density during exposure is related to dose rate, as was seen for MD-55 film. (7) It is likely that the OD increase between intra- and inter-exposure are indistinguishable at very low dose rates, where the increase on short time scale looks similar to noise, and therefore a separate radiation detector would be required to signal the end of exposure. This would complicate the dosimetry process, and the limiting dose rate may need to be established if the radiation sensitive medium of EBT film is to be used as a real-time in vivo dosimeter. The inset

84 in Figure 26 illustrates OD versus time for five pieces of film, each exposed to 9.52 Gy at 2.86 Gy/min. The measurements obtained for the same average dose rate is reproducible, yielding a 1.0% standard deviation in OD at the end of exposure. This 1.0% deviation includes errors due to possible spatial variation in film response, which may be similar to that observed in MD- 55, (11) measurement errors in the spectrophotometer results, and random light source intensity fluctuations. It is interesting that OD is slightly non-linear with time (and hence with dose). This nonlinearity with dose is not unique to the real-time measurements reported here and can also be seen in the data provided by ISP, which were measured 1 h after exposure, as well in a recent paper by Devic and colleagues. (12) It is not due to a shift in wavelength of maximum absorbance (λ max ), which was found to be quite stable over the entire 200-second exposure to 9.52 Gy (Figure 27). The average λ max for all exposures was calculated to be ± 0.7 nm (where the reported uncertainty is two times the standard deviation). The effect of changing the wavelength range of the spectral averaging window on OD is shown in Figure 28, illustrating that increasing the spectral range from 10 nm to 70 nm still showed a similar non-linear effect, albeit less pronounced. This was confirmed by Todorovic and colleagues in 2006, showing that grayscale MV 9.52 Gy 2.86 Gy/min 68 OD OD IRRADIATION Time (s) Time (s) Figure 26. Optical density versus time for a single piece of EBT film (beam on at 0 sec, off at ~ 200 sec); optical density versus time for five pieces of EBT film shown on a reduced time scale (inset).

85 OD values also exhibited non-linearity, and a decreased signal per dose. (13) One possibility is that the non-linearity is due to saturation of carbon-carbon double bonds in the polymer. (The sensitive medium of GafChromic EBT film is a modified version of that used in GafChromic MD-55 and HS films. 69 The optical density of these films is based on the increase in number of chains with conjugated double and triple bonds that form during radiation induced polymerization. (7) ) As the dose increases, the number of monomer chains that were converted to polymer chains also increases, and there are not enough reactive species (undisturbed monomer chains) present to give the same signal per dose. This is seen as decrease in sensitivity and non-linearity with dose. It is unlikely that the non-linearity can be attributed to decrease in polymerization rate with dose, since non-real time measurements also show the same trend. (12,13) The OD versus dose non-linearity is not an issue for regular dosimetric use of this film, as a simple calibration plot is all that is required. The sensitive medium of this film may also be used for real-time in vivo dosimetry, provided that the total delivered dose is kept track of and the curves are insensitive to fluctuations in dose-rate and temperature that are typical of in vivo conditions. If that s the case, a simple correction function would be needed to obtain a one- Wavelength of Maximum ΔA (nm) MV 9.52 Gy 2.86 Gy/min Time (s) Figure 27. Wavelength of maximum absorbance for EBT film versus time during and after exposure to 9.52 Gy at 2.86 Gy/min with 6 MV X-rays (beam on sec). D. Lewis, ISP (private communications, 2004 and 2005).

86 MV 9.52 Gy 2.86 Gy/min OD nm nm nm nm nm nm nm Time (s) Figure 28. Optical denisty of EBT film versus time for various spectral averaging windows. to-one correlation between OD and dose that would still allow real-time dosimetry. Otherwise, an appropriate model describing the polymerization kinetics, that can account and correct for all such variations, is desirable. A full understanding of chemistry and kinetics would require information on structure and packing of monomer units within the sensitive medium. The other possibility for the non-linearity of OD with dose is a change in the optical properties of the sensitive medium used in EBT film with dose. No information regarding these optical properties currently exists, and little can be deduced from a recent study of MD-55 film by Fusi et al., (14) since the sensitive media in MD-55 and EBT films are different (see Sec. III.D). B. Sensitivity and Stability Comparison Between EBT and MD-55 Films The OD versus time for the two types of films due to a 1.9 Gy exposure with 6 MV X- rays is shown in Figure 29. The measurements were not normalized by thickness of radiation sensitive layer (each of EBT s radiation sensitive layers is 1 μm thicker than that of MD-55). Comparing or correcting for the thickness difference between MD-55 and EBT films is not particularly relevant; the two films have different radiation sensitive media, and their suspensions (both concentration and distribution) within the active layers are likely not the same.

87 71 However, if a dosimeter were to be designed for a specific thickness, it would be useful in predicting the performance that could be achieved with each of these media. Results show the EBT film to be 7.6 ± 0.2 times (as measured immediately at the end of exposure) more sensitive than MD-55 film when exposed to approximately 2 Gy with 6 MV X- rays. The sensitivity increase is defined here as the ratio of the optical densities for a given dose, although other definitions (such as the ratio of doses required to achieve a certain optical density) can be used and may give slightly different sensitivity value. Any discussion of sensitivity between MD-55 and EBT film depends on the energy at which the exposures to radiation are performed, since EBT is supposed to be more sensitive to low kv energy photons than MD-55 (ISP product information). This makes EBT s response to ionizing radiation closer to that of water, potentially broadening the energy range over which it can be used as an in vivo dosimeter. Furthermore, since optical density of EBT (measured near the absorbance peak) is not linear with dose, increase in sensitivity between EBT and MD-55 is also dose dependent for the definition used in this report MV 1.90 Gy 2.86 Gy/min EBT MD OD Time (h) Figure 29. Optical density for EBT and MD-55 films during and after exposure.

88 72 Percent Increase in OD After End of Exposure MV 1.90 Gy 2.86 Gy/min EBT MD Time (h) Figure 30. Percent increase in OD for EBT and MD-55 films after exposure, calculated with respect to OD at the end of exposure. The percent increase in OD following exposure, calculated with respect to the value measured immediately at the end of exposure from the data of Figure 29, is shown in Figure 30. The percent increase for EBT film is nearly two times less than that of MD-55, and the signal begins to plateau earlier. At 18 hours after exposure, the OD of EBT film has increased by roughly 12.5%, whereas the increase for MD-55 film is more than 25%. The decreased postexposure darkening of EBT film, compared to that of MD-55 film, suggests that the polymerization kinetics occurring within the radiation sensitive medium are quicker, likely due to a different structure and packing of the sensitive material within the radiation sensitive layer (see Section 3.4). This is important for both conventional uses of these films and for the realtime dosimetry objectives in this work. Faster polymerization kinetics will allow accurate dose measurements soon after exposure without need for consideration of potential errors introduced by post-exposure darkening. Similarly, accurate real-time dose measurements during the latter

89 Percent Increase in OD After End of Exposure 16 6 MV 1.90 Gy Gy/min EBT MD Time (h) Figure 31. Percent increase in OD for EBT and MD-55 films within one hour after exposure. part of exposure will be possible without extraneous errors due to post-exposure darkening of the dose delivered at the onset of exposure. The percent increase in OD reported here for MD-55 film are higher than was previously reported, (15,16) and will likely be higher than will be reported for EBT film. The reason is that other measurements compare the increase in optical density to the value measured soon after exposure, but not immediately after exposure as they often use a regular non-real-time densitometer. This data shows that a large portion of growth in OD occurs within the first few minutes (Figure 31), and is thus not assessed in more conventional methods of measuring optical density for these films. It should be noted that the OD drift values reported here are valid for the dose-rate of 2.86 Gy/min, and may differ if another dose or dose-rate is used. (3) C. Dependence of Real-Time ΔOD Measurements on Dose Rate for the EBT Film The change in OD versus time for a 9.52 Gy exposure delivered at two different dose

90 MV 9.52 Gy 0.95 Gy/min 5.71 Gy/min OD OD Gy/min Gy/min Dose (Gy) Time (s) Figure 32. Optical density for EBT film exposed to 9.52 Gy, delivered with 6 MV at 0.95 Gy/min and 5.71 Gy/min; OD normalized to dose during same exposure to 9.52 Gy (inset). rates (0.95 Gy/min and 5.71 Gy/min) is shown in Figure 32. If the film is used conventionally by measuring optical density some time after the end of exposure, the six-fold dose rate variation does not appear to introduce error, although further investigations to confirm this should be performed. With time axis converted to dose, the data obtained during exposure superimpose well (inset). However, analysis of variance of the ΔOD values as measured at the end of exposure revealed a statistical difference between the two groups, and hence a dose rate dependence for real-time measurements. Again, as in previous experiments, the films were taken from the sheet at random, and irradiated at random dose rates to eliminate any systematic error that may arise from film non-uniformity or LINAC output variation. The spectrometer and LED were also checked for linearity and stability. The percent standard deviation (with respect to the mean) increased from 0.9% to 1.8% when both sets of data were included. Depending on the desired accuracy of dose estimate, this small dose rate dependence of real-time measurements may be tolerable. However, for the purposes of real-time in vivo dosimetry, dose and dose rate ranges over which dose rate variations do not introduce superfluous errors will need to be

91 75 established. The ΔOD versus dose curve can be then used up to the established dose without correcting for dose rate dependent fluctuations. The dose rates used in this investigation (0.95 Gy/min and 5.71 Gy/min) are applicable to external beam (linear accelerator output range of MU/min, with treatments typically delivered at 300 MU/min), and high dose rate brachytherapy. (17) To investigate whether this radiochromic medium is appropriate for low dose rate brachytherapy, dose rates down to a few cgy per hour will need to be used. (18) D. Structure of Active Crystals in MD-55 and EBT Films Information about structure of a chemical compound can yield insight into its function. The microscope images of the sensitive media used in MD-55 and EBT films are shown in Figure 33 (a 20 micron bar in bottom right corner indicates the scale). The monomer crystals used in MD-55 are sand-like (top image), and look similar as previously shown using scanning electron microscopy. (2) The monomer crystals within EBT, on the other hand, are elongated, stick-like structures (bottom image). The active component used in EBT film is a modified version of that used in MD-55. * Both self-assemble into the macroscopically different structures, without any steps taken to affect the macroscopic appearance of the crystals (by, for example, mechanical treatment of the crystals). Using data showing crystal structures of several monomer assemblies, a method of predicting polymerizability has been developed, based on three-dimensional packing of diacetylenes within the crystal, where the packing is believed to depend on the type and size of side groups. (19,20) It is thus reasonable to assume that the observed difference in the crystal structure between MD-55 an EBT is due to this type of modification. The difference in the crystal structure, observed visually, may be responsible for the difference in dosimetric behaviour, such as sensitivity (signal per dose) to radiation. For example, the longer crystal in EBT may be more likely to interact with ionizing radiation, inducing polymerization within the chain, compared to the shorter crystal seen in MD-55. The arrangement of formed polymer may also be different in the two crystals. For MD-55 film, the polymerization reaction proceeds in only one direction with respect to the crystal axes, (21) and the absorbance is due to the conjugated double and triple bonds within the polymer chain. (22) The change in crystal structure with increased dose, evident through a shift in absorbance peak, (21,23) increases the separation between the last polymer unit and the next available monomer, decreasing the rate of polymerization. (7), While a shift in λ max is seen for MD-55, (1,5,15) no such shift in λ max is seen for EBT film * D. Lewis, ISP (private communications, 2004 and 2005).

92 76 Figure 33. Microscope images of monomer crystals within the sensitive media of MD-55 (top) and EBT (bottom) films. (The blurring on the edges of the image is typical for an uneven thickness of sample).

93 77 (Figure 27). This suggests that the internal packing of the monomers within the active crystals of EBT may be more stable, and that little rearrangement or separation occurs between the monomer and polymer structures. Thus no significant decrease in polymerization rate is xpected to be seen. The decreased post-exposure development observed for EBT film suggests that most of the polymerization occurs during exposure, and hence the polymerization rate does not in fact decrease with dose as much as for MD-55 film. The dosimetric properties are discussed in greater detail in Chapter 7, linking the properties to physical parameters of the two radiochormic media. IV. Conclusion The new GafChromic EBT film is more sensitive to ionizing radiation dose and exhibits less post-exposure darkening than GafChromic MD-55 film. It shows a distinction between interand intra-exposure rates of ΔOD increase, which allows easy identification of the end of exposure, but ΔOD during exposure is not linear with dose. Since λ max is stable over the entire exposure to 9.52 Gy, increasing the spectral window of interest does not correct non-linearity. This non-linearity is thus likely due to the actual polymer chemistry that occurs in the sensitive medium of the film (such as depletion of monomers available for polymerization). It is not necessarily a deterrent to real-time dosimetry, as long as the correction function is stable against dose rate and temperature fluctuations. Under these conditions, a correction for this can be made using an appropriate model that can sufficiently describe the chemistry and kinetics during exposure. Developing such a model requires knowledge of monomer structure and packing of monomers within the sensitive medium. A small dose rate dependence in real-time measurements, yielding an increase in standard deviation from 0.9% to 1.8%, is seen for a dose of 9.52 Gy and a dose rate range between 0.95 and 5.71 Gy/min. For EBT film to be useful in real-time in vivo dosimetry, a dose and dose rate ranges, at which dose rate dependent errors are not seen, need to be established. Dependence of ΔOD measurements and λ max on temperature will also need to be investigated. Although further investigations are required, GafChromic EBT s increase in sensitivity and rates of polymerization make it a promising medium for realtime in vivo dosimetry.

94 ACKNOWLEDGEMENTS 78 The authors wish to thank Joanne Kniaz and Advanced Optical Microscopy Facility for their contribution. This work was in part funded by National Institutes of Health / National Institute on Aging (R21/R33 AG19381) and by the Fidani Center for Radiation Physics. 1. McLaughlin W.L., Yun-Dong Chen, Soares C.G., Miller A., Van Dyk G., Lewis D.F. Sensitometry of the response of a new radiochromic film dosimeter to gamma radiation and electron beams. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 302: , Klassen N.V., van der Zwan L., Cygler J. GafChromic MD-55: Investigated as a precision dosimeter. Medical Physics 24: , Ali I., Costescu C., Vicic M., Dempsey J.F., Williamson J.F. Dependence of radiochromic film optical density post-exposure kinetics on dose and dose fractionation. Medical Physics 30: , Ramani R., Lightstone A.W., Mason D.L.D., O'Brien P.F. The use of radiochromic film in treatment verification of dynamic stereotactic radiosurgery. Medical Physics 21: , Mack A., Mack G., Weltz D., Scheib S.G., Böttcher H.D., Seifert V. High precision film dosimetry with GafChromic films for quality assurance especially when using small fields. Medical Physics 30: , Niroomand-Rad A., Blackwell C.R., Coursey B.M., Gall K.P., Galvin J.M., McLaughlin W.L., Meigooni A.S., Nath R., Rodgers J.E., Soares C.G. Radiochromic film dosimetry: Recommendations of AAPM Radiation Therapy Committee Task Group 55. Medical Physics 25: , Rink A., Vitkin I.A., Jaffray D. Suitability of radiochromic medium for real-time optical measurements of ionizing radiation dose. Medical Physics 32: , McLaughlin W.L., Puhl J.M., Al-Sheikhly M., Christou C.A., Miller A., Kovacs A., Wojnarovits L., Lewis D.F. Novel radiochromic films for clinical dosimetry. Radiation Protection and Dosimetry 66: 263-8, 1996.

95 79 9. Milton J.S., Arnold J.C. Introduction to Probability and Statistics: principles and applications for engineering and the computing sciences. 2 ed. Toronto: McGraw-Hill, Inc., Cheung T., Butson M.J., Yu P.K.N. Use of multiple layers of GafChromic film to increase sensitivity. Physics in Medicine & Biology 46: N235-N240, Meigooni A.S., Sanders M.F., Ibbott G.S., Szeglin S.R. Dosimetric characteristics of an improved radiochromic film. Medical Physics 23: , Devic S., Seuntjens J., Sham E., Podgorsak E.B., Schmidtlein C.R., Kirov A.S., Soares C.G. Precise radiochromic film dosimetry using a flat-bed document scanner. Medical Physics 32: , Todorovic M., Fischer M., Cremers F., Thom E., Schmidt R. Evaluation of GafChromic EBT prototype B for external beam dose verification. Medical Physics 33: , Fusi F., Mercatelli L., Marconi G., Cuttone G., Romano G. Optical characterization of a radiochromic film by total reflectance and transmittance measurements. Medical Physics 31: , Chu R.D.H., Van Dyk G., Lewis D.F., O'Hara K.P.J., Buckland B.W., Dinelle F. GafChromic dosimetry media: A new high dose, thin film routine dosimeter and dose mapping tool. Radiation Physics and Chemistry 35: , Reinstein L.E., Gluckman G.R., Meek A.G. A rapid colour stabilization technique for radiochromic film dosimetry. Physics in Medicine & Biology 43: , Nath R. Physical properties and clinical uses of brachytherapy radionuclide. In: Williamson J.F., Thomadsen B.R., Nath R., eds. Brachytherapy Physics: AAPM Summer School Madison: Medical Physics Publishing Corportation, p. 7-37, Hall Eric J. Radiobiology for the Radiologist. 4 ed. Philadelphia: J.B. Lippincott Company, p. 121, Baughman R.H. Solid-state synthesis of large polymer single crystals. Journal of Polymer Science: Polymer Physics Edition 12: , 1974.

96 20. Baughman R.H., Yee K.C. Solid-state polymerization of linear and cyclic acetylenes. Journal of Polymer Science: Macromolecular Reviews 13: , Guillet J. Photopolymerization. In: Polymer Photophysics and Photochemistry: an introduction to the study of photoprocesses in marcomolecules. New York: Cambrdige University Press, p , Sixl H., Warta R. Excitons and Polarons in Polyconjugated Diacetylene Molecules. In: Kuzmany H., Mehring M., Roth S., eds. Electronic Properties of Polymers and Related Compounds. Berlin: Springer Verlag, p , Tsibouklis J., Pearson C., Song Y.P., Warren J., Petty M., Yarwood J., Petty M.C., Feast W.J. Pentacosa-10,12-diynoic Acid/Henicosa-2,4-diynylamine alternate-layer Langmuir- Blodgett films: synthesis, polymerization and electrical properties. Journal of Materials Chemistry 3: , 1993.

97 CHAPTER 4: EFFECTS OF VARYING DOSE RATE ON REAL- TIME MEASUREMENTS OF OPTICAL DENSITY OF GAFCHROMIC EBT Portions of the following have been published as Intra-irradiation Changes in Signal of Polymer Based Dosimeter (GAFCHROMIC EBT) Due to Dose Rate Variations by Alexandra Rink, I. Alex Vitkin, and David A. Jaffray in Physics in Medicine and Biology 52(22), p. N523-N529 (2007) (journal homepage at abstract at 81

98 82 I. Introduction In recent years, use of radiation-induced polymerization materials for measurement of dose has been gaining importance. (1-7) While many of the current polymer systems are utilized for two- or three-dimensional dose representation, their spatial resolution offer potential for successful implementation in point-based measurements. Radiochromic materials commercially available in GAFCHROMIC films (International Specialty Products, Wayne, NJ), which polymerize upon irradiation, have been also shown to respond quickly enough to see the effects during irradiation, or in real-time. (8,9) These radiochromic media are able to provide sufficient signal from a sub-cubic millimeter volume, (8,9) and can be made relatively energy independent, as was done with GAFCHROMIC EBT. (10,11) Ultimately, all of these characteristics may allow the use of these polymer systems in real-time dose measurements on or within the patient during diagnosis, treatment and monitoring of a patient s pathologies. However, these polymer systems are known to have post-exposure development (8,9,12,13) due to non-instantaneous polymerization reaction. For this reason, variations in real-time optical density (OD), which is used as an indicator of delivered dose, are expected with fluctuating dose rates (and thus total time taken to deliver a given dose). The impact of this dependence on dose rate needs to be quantified in order to properly assess the suitability of these radiochromic materials for a real-time dosimetry. The (8) signal measured for MD-55 film was shown to depend on dose rate for doses above 1 Gy. For EBT film, it is expected that the dose rate dependence is going to be less significant than that of its predecessor since previous results suggest that polymerization reactions in EBT film occur faster than in MD-55 film. (9) This note quantifies the dependence of real-time change in OD of EBT film on the rate of dose delivery. II. Methods and Materials 1 cm 1 cm pieces of GAFCHROMIC EBT (Lot I) film were placed into the film holder within a 30 cm 30 cm 4 cm phantom, (11) such that the centre of the film was located at 1.5 cm depth, and the plane of the film is perpendicular to the top surface of the phantom and along the central axis of the irradiating beam. The phantom was placed onto 30 cm 30 cm 6 cm slab of Solid Water. A range of doses ( cgy dose to water at the centre of film) was then delivered at various dose-rates ( cgy/min) using a 100 cm SAD (98.5 cm SSD) setup, 10 cm 10 cm field, and 6 MV X-rays from a linear accelerator (Elekta Synergy ).

99 83 The details of the delivery and collection optical fibers are described elsewhere. (11) In brief, a 50/125 μm (core/cladding diameter) optical fiber delivered interrogation light from the 630-nm light emitting diode (LED5 23RED, 17 nm full-width half-maximum, LED Light Inc., Carson City, NV) perpendicularly to the film. A 1.50/1.55 mm fiber on the other side of the film collected the transmitted light and delivered it to the spectrophotometer (SD2000, Ocean Optics Inc., Dunedin, FL). Both delivery and collection fibers were located at 1.5 cm depth within the phantom, parallel to the top surface, and perpendicular to the EBT film. The spectra of transmitted light were obtained for approximately 10 seconds prior to exposure, during exposure, and for several minutes after exposure at roughly one to two spectra per second. Using Matlab 7.1 for all data processing and analysis, the absorbance was calculated at each wavelength using equation 1: ( I R ( λ) I D ( λ)) Δ A( λ) = log10 (1) ( I ( λ) I ( λ)) S where I D is the dark spectrum, I R is the reference spectrum (light transmitted through film prior to irradiation), and I S is the sample spectrum at every measurement prior, during and after irradiation. Optical density for each collected spectrum was then calculated using a nm wavelength range centered on the main absorbance peak using equation 2: OD = λ n 1 1 n λ1 i= 1 D ΔA( λi ) + ΔA( λ ( 2 i+ 1 ) )( λ i+ 1 λ ) where ΔA(λ i ) is the absorbance calculated at wavelength λ i within the 10 nm range. (8,9) i (2) The OD was then plotted as a function of time. Straight lines were fitted to pre-exposure and postexposure data; a third order polynomial was fitted to the data obtained during exposure. Using the intercepts between these lines, and subtracting the OD before from the OD at the end of exposure (an average of five points each time), the real-time ΔOD for the given dose was calculated as illustrated in Figure 34. (8) For most of the doses and dose rates, 5 measurements were made using 5 different 1 cm 1 cm pieces of EBT film from the same sheet. The exception is in the cases where only three or four measurements were made for the following doses and dose rates: 5 cgy delivered at 130 and 260 cgy/min, 10 cgy delivered at 260 and 390 cgy/min and 25 cgy delivered at 520 cgy/min. No data for the lowest dose of 5 cgy at 390 and 520 cgy/min or 10 cgy at 520 cgy/min was available. This was due to a very short irradiation time and not enough data points available during irradiation to do a line of best fit to calculate ΔOD (as described above).

100 84 Figure 34. Optical density versus time for a 50 cgy irradiation at 16 cgy/min. The ΔOD for this dose and dose rate is calculated as the difference between the ΔOD at the end of exposure and that prior to exposure. The same approach is used for other doses and dose rates. An average ΔOD, or <ΔOD Dose >, and standard deviation, σ OD, for each dose and dose rate were calculated using equations 3 and 4, where n is the total number of measurements using different pieces of film. These were then normalized to 1 Gy in order to condense the scale for easier comparison, and plotted for each dose as a function of dose rate. n < Δ >= Δ ODi OD Dose (3) n n i 2 ( ΔODi < ΔODDose > ) σ OD i = ( n 1) (4) The average percent standard deviation for each dose was then calculated across all dose rates, and reported as <%σ OD >. For example, for 100 cgy the percent standard deviations of ΔOD at each dose rate used are 1.5%, 1.5%, 1.7%, 2.0%, 1.3%, 1.4% and 1.3%. The average

101 85 percent standard deviation, <%σ OD >, for 100 cgy is then 1.5%. To assess dose-rate response, the ΔOD is averaged across all film samples, and the overall percent standard deviation for the given dose using all dose rates is calculated, reported as %σ OD. This number incorporates the deviations in σod arising from using different dose rates, and is always larger than <%σ OD >. Since dose rate dependent systematic error contributes to the %σod, it can also be thought of as 1σ uncertainty in measurement if the dose rate (somewhere in cgy/min) is unknown. The difference between the two, %σ OD -<%σ OD >, denoted as Δ, is the average increase in uncertainty when the dose rate is varied between 16 cgy/min and 520 cgy/min instead of keeping it constant. III. Results and Discussion Figure 35. The average sensitivity ΔOD/D Gy as a function of dose rate, for various doses (error bars are ±σ OD /D Gy, where σ OD is one standard deviation for the ΔOD obtained for that dose and given dose rate and D Gy is the delivered dose). The lines of best fit are shown on the graph. Because of overlap between some of these, the lines of best fit for 5 cgy and 10 cgy doses are marked as dashed lines for clarity.

102 86 Average ΔOD normalized to 1 Gy, denoted as ΔOD/D Gy, is plotted versus dose rate for various doses (Figure 35). First it can be seen that the ΔOD/D Gy decreases with dose, dropping from ~0.6/Gy for 5-10 cgy to just over 0.2/Gy for 1000 cgy. This is due to previously reported nonlinearity of the EBT film with dose. (9,10,14) It can also be seen that as the dose rate increases, the ΔOD/D Gy for a given dose decreases, and vice versa. This behaviour can be explained by the fact that at low dose rates, the polymerization reactions initiated at the beginning of irradiation have had more time to develop, and hence the ΔOD and ΔOD/D Gy associated with the formed polymers is higher than if the dose were delivered at a high dose rate, and thus over a shorter irradiation time. Although the decrease due to higher dose rate appears relatively minor, a careful analysis of the data is warranted in order to estimate the importance of this decrease in ΔOD. The linear fits to the decrease in ΔOD/D Gy with dose rate are also plotted, and the coefficients of these lines of best fit are shown in Table 4. The intercept, y, is the ΔOD/D Gy value approached as dose rate approaches zero (in other words, as the time for a given dose approaches infinity). The idea that as time approaches infinity, the ΔOD for a given dose converges to a given value has been reported previously for various fractionation patterns. (15) Our data support that, since a fractionation pattern can be thought of as a dose rate variation pattern, just on a longer time scale. The slope, m, of the lines of best fit appears to decrease with increasing dose. However, with the exception of the data for 5 cgy, it does so proportionally with the intercept and the overall sensitivity of the film. Thus deducing that the dose rate fluctuations have a smaller effect on larger doses from this data alone would be erroneous. If there were no fluctuations associated with the ΔOD readout for a given dose, then the percent decrease in ΔOD/D Gy and in ΔOD caused by increasing the dose rate from 16 cgy/min to 520 cgy/min calculated using the best-fit lines seem rather substantial. As shown in Table 4, it rangers between 4.5 and 6.7%, with no obvious dose-level trend. However, the film is known to vary in sensitivity with a deviation of ±1.5% to ±4% depending on dose. (14) Also, small fluctuations in light source output and spectrophotometer (measured combined standard deviation of 0.1% over a period of 5 hours) exist and the linear accelerator output has a maximum ±2% variation. Thus, the ΔOD measurements are not error free, and the important question is whether the error introduced due to fluctuating dose rate is a significant portion of the overall error associated with the ΔOD measurement.

103 87 Table 4. Coefficients of equations of best fit characterizing ΔOD/D Gy as a function of dose rate ( D & ). N/A denotes data that are not available. y denotes the ΔOD/D Gy as t, regardless of the dose rate used for irradiation. Dose (cgy) equation of best fit line: % decrease in Fit of data in ( ΔOD / D Gy = y + m D ) y m ( 10-3 ) (1/Gy) (min/gy 2 ) ΔOD from 16 to Figure 35 to line 520 cgy/min rate of best fit (R 2 ) ± ± 4.5 N/A ± ± 3.0 N/A ± ± ± ± ± ± ± ± ± ± ± ± To evaluate whether dose rate has a statistically significant effect on the ΔOD measurements, an F-test (16) was performed for each dose and different combinations of dose rates, starting with only two groups (16 and 32 cgy/min), and ending with seven groups (16 to 520 cgy/min) if data were available. This test compares the difference between the mean ΔOD values of groups with respect to the standard deviation of each group, in order to detect whether the difference between groups is significant compared to the associated uncertainty in measurement. If the difference between means of groups is deemed to be statistically significant, then varying the dose rate introduces a systematic error on ΔOD measurement, increasing the overall uncertainty of the measurement if the dose rate is not known. Since as many as five pieces of film (taken randomly from a single large piece of EBT film) were used at each dose and dose rate to obtain these average ΔOD and σ OD values, the uncertainty associated with ΔOD measurement will include film non-uniformity, variability in output of the linear accelerator between different times that the measurements were performed, and light and spectrophotometer fluctuations. If p<0.05, the difference between the sets of measurements is taken to be statistically significant, and thus dose rate fluctuations have a significant effect on the ΔOD

104 88 measurement. For 5 cgy, varying the dose rate (even between 16 cgy/min and 32 cgy/min) introduced significant difference in ΔOD beyond normal deviation. For all other doses, varying the dose rate from 16 cgy/min to 390 cgy/min (or 520 cgy/min) introduced significant change in ΔOD. The general trend is that p<0.05 levels are seen when dose rate difference is eightfold or more for a given dose. For example, there is a statistically significant difference between ΔOD values obtained at 16 and 130 cgy/min, or between 520 and 65 cgy/min, but not between 520 and 390 cgy/min, or 16 and 65 cgy/min. This seems to be irrespective of the dose delivered to the film. Table 5. Average percent standard deviation (<%σ OD >) for each dose, 1σ uncertainty (%σ OD ), and the difference between the two (Δ) for each dose delivered. Dose (cgy) <%σ OD > %σ OD Δ Percent standard deviations of ΔOD were calculated for a given dose at specific dose rates, and the average of these are shown as <%σ OD > in Table 5. Overall per cent standard deviation, or 1σ uncertainty when dose rate is unknown, is also reported for each dose. A small increase in uncertainty is observed when ΔOD values obtained with various dose rates across the entire range tested are combined. This further illustrates the dependence of measured ΔOD on the dose rate used. For all doses, except 5 and 10 cgy, the average increase in uncertainty, denoted as Δ, is ~1%. For all doses examined in the 5 to 1000 cgy range, the %σ OD calculated using ΔOD values obtained with the entire in 16 to 520 cgy/min range is 4.4% or less. IV. Conclusion Real-time measurements of ΔOD of GAFCHROMIC EBT film irradiated to doses in the 5 cgy to 1000 cgy range showed a small dose rate dependence when an eightfold difference in dose rate was introduced. Combining all ΔOD measurements for a given dose irrespective of dose rate used within the cgy/min range introduced an increase in measurement uncertainty of

105 89 ~1% on top of the already present uncertainty associated with random errors of measurement. This increase in uncertainty is due to systematic errors associated with dose rate variability. For all doses, the combined uncertainty in ΔOD is <4.5% for some unknown dose rate in the cgy/min range. Therefore, although varying dose rate by an order of magnitude has a statistically significant effect on the real-time ΔOD, the measurement can still be performed with an uncertainty of 4.5% or less, which can be satisfactory for many applications of EBT film. ACKNOWLEDGEMENTS This work was in part funded by Research Studentship of the Terry Fox Foundation through an award (016646) from the National Cancer Institute of Canada, Scace Prostate Cancer Award, National Institutes of Health / National Institute on Aging (R21/R33 AG19381), and by the Fidani Center for Radiation Physics. 1. Maryanski M.J., Ibbott G.S., Eastman P., Schulz R.J., Gore J.C. Radiation therapy dosimetry using magnetic resonance imaging of polymer gels. 9Medical Physics 23: , Low D.A., Dempsey J.F., Venkatesan R., Mutic S., Markman J., Haacke E.M., Purdy J.A. Evaluation of polymer gels and MRI as a 3-D dosimeter for intensity modulated radiation therapy. Medical Physics 26: , Berg A., Ertl A., Moser E. High resolution polymer gel dosimetry by parameter selective MR-microimaging on a whole body scanner at 3 T. Medical Physics 28: , Wuu C.S., Schiff P., Maryanski M.J., Liu T., Borzillary S., Weinberger J. Dosimetry study of Re-188 liquid balloon for intravascular brachytherapy using polymer gel dosimeters and laser-beam optical CT scanner. Medical Physics 30: 132-7, Oldham M., Siewerdsen J.H., Kumar S., Wong J., Jaffray D.A. Optical-CT gel dosimetry I: Basic investigations. Medical Physics 30: , Chiu-Tsao S., Duckworth T.L., Patel N.S., Pisch J., Harrison L.B. Verification of Ir-192 near source dosimetry using GAFCHROMIC film. Medical Physics 31: 201-7, 2004.

106 90 7. Hirata E.Y., Cunningham C., Micka J.A., Keller H., Kissick M.W., DeWerd L.A. Low dose fraction behavior of high sensitivity radiochromic film. Medical Physics 32: , Rink A., Vitkin I.A., Jaffray D. Suitability of radiochromic medium for real-time optical measurements of ionizing radiation dose. Medical Physics 32: , Rink A., Vitkin I.A., Jaffray D.A. Characterization and real-time optical measurements of the ionizing radiation dose response for a new radiochromic medium. Medical Physics 32: , Chiu-Tsao S., Ho Y., Shankar R., Wang L., Harrison L.B. Energy dependence of response of new high sensitivity radiochromic films for megavoltage and kilovoltage radiation energies. Medical Physics 32: , Rink A., Vitkin I.A., Jaffray D.A. Energy dependence (75 kvp to 18 MV) of radiochromic films assessed using a real-time optical dosimeter. Medical Physics 34: , Klassen N.V., van der Zwan L., Cygler J. GafChromic MD-55: Investigated as a precision dosimeter. Medical Physics 24: , Ali I., Costescu C., Vicic M., Dempsey J.F., Williamson J.F. Dependence of radiochromic film optical density post-exposure kinetics on dose and dose fractionation. Medical Physics 30: , Zeidan O.A., Li J.G., Low D.A., Dempsey J.F. Comparison of small photon beams measured using radiochromic and silver-halide films in solid water phantoms. Medical Physics 31: , Ali I., Williamson J.F., Costescu C., Dempsey J.F. Dependence of radiochromic film response kinetics on fractionated doses. Applied Radiation and Isotopes 62: , Milton J.S., Arnold J.C. Introduction to Probability and Statistics: principles and applications for engineering and the computing sciences. 2 ed. Toronto: McGraw-Hill, Inc., 1990.

107 CHAPTER 5: CHARACTERIZATION OF GAFCHROMIC EBT: TEMPERATURE AND HUMIDITY EFFECTS Written in collaboration with Dr. D.F. Lewis and Dr. S. Varma of International Specialty Products, to be submitted for publication as Temperature and Hydration Effects on Absorbance Spectra and Sensitivity to Ionizing Radiation of a Radiochromic Medium by Alexandra Rink, David F. Lewis, Sangya Varma, I. Alex Vitkin, and David A. Jaffray 91

108 92 I. Introduction Various polymer systems have been investigated for applicability as ionizing radiation dosimeters. (1-9) However, before implementing any polymer system as a patient dosimeter, the underlying reactions and the effect of dose, dose rate, temperature, and humidity variations on the dose estimate should be understood. Chemical Background and General Experience The active component in MD-55 and EBT films are pentacosa-10,12-diynoic acid (PCDA) and the lithium salt of pentacosa-10,12-diynoic acid (or LiPCDA), respectively. They are from a class of materials known as diacetylenes, and the chemical formulas are illustrated in Figure 36. These organic molecules contain two conjugated acetylene groups (carbon-carbon triple bonds). When in an ordered state, for example in crystals, monomolecular layers or micelles, some diacetylenes undergo a 1,4 polymerization reaction initiated by exposure to ionizing radiation. The resulting diacetylene polymers are intensely colored materials produced in proportion to the absorbed dose. It has been shown that for polymerization of diacetylenes similar to that used in MD-55, the diacetylene monomers should be packed so that the triple bonds in adjacent monomers are within 0.4 nm. (10) The efficiency with which the polymerization proceeds is related to the intermolecular proximity and relative orientation of the diacetylene monomer molecules. It is thought to be likely that the molecular orientation and proximity factors leading to higher or lower sensitivity will be similar for the EBT and MD-55 films. The packing of the monomers is believed to depend on the type and size of diacetylene side groups. (11,12) While the molecular packing of the active diacetylene monomer in EBT film is likely not the same as that of MD-55 film, since they self-assemble into visually very different structures, (9) both contain the pentacosa-10,12-diynoate ion (Figure 36, right side without Li + ). Thus, the packing of pentacosa-10,12-diynoate ions is also dependent on the size and chemistry of the cation, which is part of one of the side groups, and this leads to dramatic differences in sensitivity. The sodium and potassium salts of pentacosa-10,12-diynoic acid (PCDA) are relatively insensitive to ionizing radiation and do not undergo significant and useful polymerization at doses less than about 1 kgy. In contrast, the parent acid (the cation is H + ) employed in MD-55 film is useful at doses greater than about 5 Gy, while the lithium salt of PCDA used in EBT film can be employed at doses as low as 1 cgy.

109 93 O OH 1 O O - Li + C10 C C C C C C C ` Figure 36. Chemical formula of pentacoasa-10,12-dyinoic acid (PCDA) on the left, and lithium salt of PCDA (LiPCDA) on the right. The 1 st and 10 th carbon atoms are marked as per nomenclature. The structure of a diacetylene polymer is often slightly different than that of the monomer chain due to tilting of each unit with polymerization. (11) Initially, the structure of the polymer is controlled by the structure of the surrounding monomer. Once the local polymer concentration increases and the surrounding monomers can no longer produce enough strain, the polymer strains will reorganize into a lower energy arrangement. (10) A transition in the structure of the diacetylene backbone is accompanied by a shift in the absorption peaks, and by a change in length of polymer chain. (13) Such a rearrangement is observed in MD-55 with increased dose (14-17) and with temperature. (8,15,18) A shift in absorption peak with dose was not observed for EBT, (9) which leads to a belief that the polymer backbone of the LiPCDA formed within EBT is not much different than the backbone of the monomer chain. Whether such a shift occurs with

110 94 temperature, and how this would affect real-time estimation of dose still remains to be determined. In preparing micro-particles of the active LiPCDA component two of the authors (Dr. D.F. Lewis and Dr. S. Varma, researchers at ISP) have discovered that two distinct forms can be grown (visually similar to Figure 25 in Chapter 3). These forms are described as hair-like and plate-like. The hair-like form describes particles that have aspect ratio (length:width) greater than 10:1, in some cases even as high as 1000:1. Depending on the aspect ratio the particles may appear literally like hairs, or, for lower aspect ratios appear more like bristles or rods. The platelike form describes particles having an aspect ratio less than 2:1 and a thickness about 10X less than the width or length. The hair-like particles, whether hairs, bristles or rods all exhibit much greater sensitivity to radiation exposure than the plate-like form, possibly due to molecular packing and configuration conducive to higher radiation sensitivity. Spectroscopic evidence that the inter-molecular structure of the two is different is presented in this Chapter. Humidity investigations of MD-55 film showed a dependence of optical density on percent relative humidity. (14) The EBT film, the more sensitive medium of the two, has not been investigated for humidity dependence. It is possible that these radiochromic media will not meet every desired criterion 8 of an ideal in vivo dosimeter. In this event, it may be important to understand the relationship of chemical composition and intermolecular packing to energy dependence, temperature dependence, and dose rate dependence. The results may provide useful feedback for help in improving dosimeter design and configuration and lead to better performance. This paper serves to investigate the dependence of the response of the EBT film upon temperature and its state of hydration. II. Methods and Materials Since the radiochromic material under investigation is intended for real-time use, during which the temperature of the surroundings (i.e. tissue) and the medium can vary, it is important to look at the temperature effects in the same manner. That is, the temperature during irradiation and during readout must be one and the same because of the simultaneity of the two. Hence, for the temperature dependence study, a real-time approach was used for measurements of ΔOD. On the other hand, humidity of the surroundings is not relevant for the intended use of the dosimeter, as the sensitive medium would be isolated from the tissue such that water molecules don t penetrate through the barrier. Thus, real-time measurements are not necessary in this case, and change in

111 MV 75kVp Change in Absorbance for 1 Gy Wavelength (nm) Figure 37. Change in absorbance spectra for EBT film exposed to 1 Gy with 6 MV and 75 kvp beams. Each spectrum is an average of five films, with error bars representative of one standard deviation in measurement (1σ) at the given wavelength. optical density as well as spectra were obtained under regular conditions some time after the end of irradiation. While care was taken to make sure the source of ionizing radiation was the same for a given experiment, this wasn t the case between the two sets of experiments. However, using a 6 MV beam for the temperature dependence study and a 150 kvp beam for the humidity dependence study shouldn t matter, as EBT was shown to be energy independent. (19) Figure 37 illustrates the average absorbance spectra obtained during irradiation to 1Gy with a 6 MV and a 75 kvp beam (data drawn from previous experiments). (19) Whether the difference between the two curves at 500 nm is significant was not investigated, but it would not affect OD measurements in the red region of the visible spectrum, and is unlikely to have a significant effect on the OD measured over the entire spectral range shown.

112 A. Temperature Dependence 96 A 30 cm 30 cm 4 cm Solid Water phantom (19) was designed to have the centre of the 1 cm 1 cm piece of radiochromic film located at 1.5 cm depth with the plane of the film perpendicular to the top surface of the phantom. It was modified to accommodate two plastic hoses (1.2 cm diameter) for water circulation (Figure 38). The water flow and water temperature were controlled with a pump (Model FSe, Haake, Germany). A 50/125 μm (core/cladding diameter) optical fiber delivered light from a 630-nm light emitting diode (LED5 23RED, 17 nm full-width half-maximum, LED Light Inc., Carson City, NV) in perpendicular incidence with the film. A 1.50/1.55 mm fiber on the opposite side of the film collected the transmitted light and delivered it to the spectrophotometer (SD2000, Ocean Optics Inc., Dunedin, FL). Both delivery and collection fibers were located at 1.5 cm depth within the phantom, parallel to the top surface, and perpendicular to the EBT film (Lot Number I). Spectra of the light transmitted by 30 cm 15 cm 1.5 cm water hoses 30 cm optical fibers 30 cm film 15 cm 1.2 cm 1.7 cm 4 cm 4 cm thermocouple 1.5 cm 30 cm 15 cm Figure 38. Schematic of the modified phantom, with plastic water hoses on either side of the film and optical fibers. The centre axes of the hoses run parallel to the optical path, 1.7 cm away, at 1.5 cm depth.

113 97 the film were obtained for approximately 10 seconds prior to the start of an exposure, during the exposure, and for several minutes after the exposure at a rate of approximately one spectrum per second. For each spectrum, an absorbance spectrum was calculated using the initial transmitted light as the background. The optical density (OD) was then calculated by integrating the absorbances between nm. The OD was plotted as a function of time. The data obtained during pre-exposure, and post-exposure intervals were fitted to linear functions; the data for the exposure interval was fitted to a third-order polynomial. Using the intercepts between these lines, and subtracting the OD before exposure from the OD at the end of exposure, the real-time ΔOD for the given dose was calculated. (8) A small hole was drilled parallel to the collection fiber to accommodate the thermocouple wire. The temperature of the film was measured using a thermocouple sensor (Fluke 179, Fluke Co., Everett, WA), positioned 4.5 mm from the centre of the film. In a separate set of measurements, two thermocouples were used to verify the equivalence of temperature at the two positions: one measured at the position of the centre of the film and the other at the regular position of the thermocouple. It was found that the temperatures of the thermocouples were always within ± 0.4 ºC, and for 75% of the time they were within ± 0.2 ºC of each other for the entire experimental range of 22 ºC to 38 ºC. For the irradiations, film was inserted into the holder at least 15 minutes prior to measurements. The temperature measured by the thermocouple 4.5 mm from the film was recorded before and after each exposure. The temperature during the exposure was taken as the average of those two measurements. For irradiation with a 6 MV beam, the phantom was positioned, without any extra buildup, under the linear accelerator at 100 cm SAD (98.5 cm SSD). A cm field at SAD was used. Each film was irradiated to a nominal 100 cgy dose (no correction for lack of backscattering material was made) at 130 cgy/min dose rate. Spectra of the light transmitted by the film were collected and used to calculate the ΔOD of the film for 100 cgy dose at the measured temperature. For part of the study, another set of films was similarly irradiated to doses 50, 100, 200 or 400 cgy. B. Absorbance and Sensitivity Dependence on Water Content For the purpose of this work, the LiPCDA was grown in two forms. The hair-like form, employed commercially as the active component in GAFCHROMIC EBT dosimetry film, was grown in a gelatin binder and coated onto a single piece of polyester. This is referred to as unlaminated film A. The second, plate-like form, which is not used commercially, was

114 98 also prepared in gelatin and coated onto polyester. It is referred to as unlaminated film B. Standard commercial GAFCHROMIC EBT film product is constructed by combining two of the unlaminated film A onto a common binding layer. Thus several pieces ( cm 2 ) of the standard EBT configuration (fully laminated film) were made from film A. Samples of this EBT film together with the unlaminated film A were then desiccated in a 50 ºC oven for 24, 48, 60 and 126 hours. The densities (weighted at each wavelength by a known response of the human eye, and termed visual densities, OD v ) of the unirradiated desiccated films were measured at ambient temperature (20-24 C) on an X-Rite 310T transmission densitometer (X-Rite Inc., Grand Rapids, MI). The films were then irradiated to a dose of 3 Gy at a rate of approximately 2 Gy/min in a Pantak 160 X-ray cabinet (Agfa, Mortsel, Belgium). They were positioned about 30 cm from the focal point of the tube and exposed to a 150 kv p 20 ma beam filtered through 2.0 mm of aluminum. Ten minutes after the end of the irradiation, the visual densities of the films were re-measured, and the change in visual density (ΔOD v ) due to the radiation exposure was calculated. The net changes in visual density of the desiccated samples were compared to the corresponding values for films which were not desiccated. A piece of unlaminated film A was also irradiated as described above and the absorbance spectrum of this film was measured with a GBC Cintra 20 UV-VIS spectrophotometer (GBC Scientific Equipment Pty Ltd., Melbourne, Australia) 30 minutes after the end of irradiation. The spectrum of unirradiated unlaminated film A was also obtained and subtracted from the irradiated film spectrum to calculate the net change in absorbance due to the radiation exposure. These two films were then desiccated in a 50 ºC oven and the spectra were re-measured after 24 and 48 hours, with the spectrum of unirradiated film used as background. Following this, the desiccated films were re-hydrated by suspending them over water in a sealed container at room temperature. The films were not in contact with liquid water. The absorbance spectra of the rehydrated unlaminated film A samples were again measured after 96 hours. In order to investigate whether an intermediate state between the regular hydrated form and the desiccated form of LiPCDA exists, a large uniform piece of film from a production roll was used. This piece of film was exposed to an arbitrary dose equivalent to ~10 Gy and then cut up into smaller pieces. All of these were placed into the same desiccator, with one piece removed at specified hours to measure the UV-VIS spectrum. No unirradiated films were used to subtract background.

115 99 Finally, the sample of unlaminated film B (made with the plate-like form of LiPCDA) was irradiated to 3 Gy and an absorbance spectrum was obtained for comparison to the spectrum of the unlaminated film A samples after desiccation and re-hydration. III. Results and Discussion A. Temperature Dependence The temperature results for commercially available EBT film is shown in the following several graphs. The wavelength of maximum change in absorbance (λ max ) as a function of temperature for films irradiated to 1 Gy is shown in Figure 39. Over the tested range of 22 ºC to 38 ºC a straight line fit to the data is given by λ = (644.7 ± 0.5) (0.41± 0.01) T ( o C) (1) max with the correlation coefficient R 2 of The trend of a decrease in λ max with increase in temperature was previously observed for other radiochromic materials. (8,15,18) This decrease appears to be independent of the dose received by the film (Figure 40), and is not unexpected since λ max was previously shown to be dose level independent. (9) The observed fluctuations in the measured λ max are possibly to be due to variation in water content within the sensitive layer, since the pieces of film are taken from the sheet randomly, and thus some measurements are performed closer to the initial sheet edge. As discussed in section III.C, water content within the sensitive layer plays a role both in the spectral absorbance peaks and in the sensitivity of the film to ionizing radiation. The other possibility is that the increase in temperature changes the three-dimensional organization of polymer, causing a shift in the λ max. Figure 41 shows the ΔOD for a dose of 1 Gy as a function of temperature. The net change in absorbance has been calculated over the nm range (ΔOD nm ), corresponding to a range of about ± 5 nm around the wavelength of maximum absorbance at room temperature. The ΔOD nm changes linearly with temperature and decreases by more than 10% over the range. The fit is described by o ΔOD = (0.631± 0.001) ( ± ) T ( ) (2) nm C with R 2 correlation coefficient of This line of best fit is later used as part of a temperature correction scheme. To show that the ΔOD decrease is not due to the observed shift in the wavelength of peak absorbance, λ max, the ΔOD at each temperature was recalculated over a 10 nm wide band centered on the observed λ max. The ΔOD 10nm still decreases (Figure 41) with

116 λ max (nm) Temperature ( o C) Figure 39. Wavelength of maximum change in absorbance of commercial EBT films irradiated to 1 Gy as a function of measured temperature. The λ max was tracked from a few seconds after the beginning of irradiation, through to its end. The error bars represent the standard deviation of λ max over this time interval cgy 100 cgy 200 cgy 400 cgy λ max (nm) Temperature ( o C) Figure 40. Values of λ max for various doses delivered to commercial EBT films as a function of measured temperature. The λ max was tracked from a few seconds after the beginning of irradiation through to the end. The error bars represent the standard deviation of λ max over this time interval.

117 ΔOD( nm) ΔOD(10nm) ΔOD Temperature ( o C) Figure 41. Change in optical density for 1 Gy dose calculated for optical range of nm, and an optical range of 10 nm centered about λ max, versus measured temperature. Line of best fit is plotted for ΔOD nm, and is later used in a temperature correction scheme. increasing temperature. However the slope, ΔOD 10nm/ ΔT, is less than that for the previous case, ΔOD / ΔT. This illustrates that the decrease in net absorbance with increasing temperature can not be attributed solely to the shift in λ max with temperature. Given the linear dependence of λ max and ΔOD nm on temperature in the ºC range, a correction scheme was designed and investigated. In this scheme the value of λ max was used to correct for the decrease in ΔOD nm measured for pieces of film irradiated to 1 Gy dose. Substituting the value of λ max in Equation (1) allows calculation of the predicted temperature, T p. A comparison between predicted and measured temperature is shown in Figure 42, together with the lines of best fit and identity. For all five measurements, the predicted temperature is within prediction error of the true value. The predicted temperature, T P, was substituted in Equation (3) and used to calculate a correction factor. This correction factor can be applied to the change in optical density at temperature T P, ΔOD TP, to determine the change in optical density at 22ºC. The correction factor is given by

118 Predicted Tempreature ( o C) Line of Best Fit Line of Identity Temperature ( o C) 102 Figure 42. Temperature calculated using the position of λ max versus measured temperature, shown with a line of best fit. The error bar is prediction error of 1σ ΔOD ΔOD( nm) 0.44 ΔOD(10nm about shifting peak) ΔOD(predicted from eq.3) Temperature ( o C) Figure 43. Change in optical density for films irradiated to 1 Gy using a ( ) fixed optical integration range of 630 to 640 nm, ( ) moving optical range of 10 nm about the peak of maximum change in absorbance, and ( ) as calculated using the peak of maximum absorbance and temperature-dependent correction factor. Error bars indicate one standard deviation or standard error (1σ), and the dashed line through the corrected ΔOD is the line of best fit.

119 103 ( a + 22b) F = (3) ( a + T b) where a and b are respectively intercept and slope of Equation (2). The change in optical density corrected to 22ºC is P ΔOD22 = F Δ (4) OD TP Figure 43 shows the effectiveness of this approach as well as the spectral range approach where the ΔOD is calculated over a wavelength range of ± 5 nm either side of the absorption maximum. The fluctuations in the ΔOD values are due to the variation in film response. If more data points were used, data similar to that in Figure 41 would be expected. While the ΔOD ( nm) decreases as seen previously (Figure 41) by about 20% as temperature increases from 24ºC to 37ºC, the change of ΔOD 10 is significantly less over the same temperature range. The approach of using λ max to determine the temperature and calculate a correction factor appears to partially remove the systematic error (the line of best fit has a small negative slope), and the standard errors of the corrected net optical densities are all less than 2.3%. The equations describing the relationships of λ max and ΔOD ( nm) to temperature are valid over the range from 22 ºC to 38 ºC for a dose of about 1 Gy. The effect of dose on the correction factors has been investigated for doses from 0.5Gy to 4Gy. The results are summarized in Table 6. Correction factors for doses of 1 Gy were determined twice. The values are shown in two columns. The first set (A) was used to correct the data shown in Figure 41. The second set of values (B) were obtained from new film samples cut from the same sheet of Table 6. Correction factors for the temperature correction scheme, calculated for doses of cgy shown for a selection of predicted temperature values. The subscript denotes the dose to which the films were exposed (two sets of films were exposed to 100 cgy, each set denoted as either A or B). T pred F 50cGy F 100cGyA F 100cGyB F 200cGy F 400cGy 25.3 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± 0.006

120 104 EBT film from which Set A was taken. From the data in Table 6 it can be seen that the correction factors for the two sets of 1 Gy irradiations are the same within experimental error. While the values of the correction factors for doses of 0.5Gy, 1Gy and 2Gy are in close correspondence and often within experimental error, the correction factors for 4Gy dose are significantly smaller. The reason for this difference is unclear and it remains to be investigated further. Two explanations for the observed negative value of ΔOD/ΔT for a given dose can be proposed. First, ΔOD decrease could be due to a decrease in sensitivity (and thus decrease in degree of polymerization). In most cases it is expected that an increase in the temperature at which irradiation occurs will result in increased sensitivity so long as the temperature is lower than that at which a phase change occurs. The increase in sensitivity relates to greater atomic motion at higher temperature and the increased likelihood that polymerization will be initiated between adjacent molecules. If the temperature increases sufficiently to cause a phase change, dramatic changes in sensitivity may result. Thus if the melting point is exceeded, the intermoleculer order of the diacetylene molecules is lost and the sensitivity diminishes to zero. While it may be theoretically possible for increasing temperature to cause a phase change leading to increased sensitivity, the authors do not know of any diacetylenes exhibiting this characteristic. The melting point of the lithium salt of PCDA used in EBT film is approximately 150 C. Therefore we believe it is unlikely that the negative change in net optical density at increasing temperatures up to 37 C is due to a decrease in the intrinsic sensitivity of the active component. B. Absorbance and Sensitivity Dependence on Water Content When films containing the lithium salt of pentacosa-10,12-diynoic acid as the active component are desiccated they undergo a significant change in sensitivity. The change in visual density (ΔOD v ) for regular (laminated) GAFCHROMIC EBT film and the unlaminated Film A are shown in Figure 44, plotted as percent loss in sensitivity compared to the ΔOD v for nondesiccated films. It can be seen that the loss of sensitivity for the unlaminated film A occurs much faster (within 24 hours), dropping by approximately 66%, but thereafter remains stable. In comparison, the sensitivity of the regular laminated EBT film decreases by roughly 25% after 24 hours. It continues to lose sensitivity at longer times, stabilizing at just over 60% decrease in sensitivity at 60 hours. In the unlaminated film the active layer is coated on one side of a polyester substrate and exposed to the environment. In the regular EBT film the active layer is

121 protected on both sides by the polyester substrate. Presumably, the laminated film loses water from the sensitive layer more slowly because moisture must diffuse through polyester or through the active layer and out the sides. The permeability of polyester to water vapor is about 1.5x10-8 cm 3.cm/cm 2.sec.kPa. This is more than two orders of magnitude less than the value of 8.6x10-6 cm 3.cm/cm 2.sec.kPa reported by Avena Bustillos et al. for mammalian gelatin. (20) 105 The active layers in EBT film are composed of about 60% gelatin and 40% active component. Since the active component is significantly more hydrophobic than gelatin the permeability of the active layer to water is expected to be less than that of gelatin alone. Because the polyester is relatively thin (97 μm), loss of water through the polyester of the film samples will not be negligible even though the permeability of polyester is much lower than gelatin The EBT samples were about 2.5cm x 2.5cm, therefore loss of moisture from the center to the sides is also slow. The distance over which a water molecule must diffuse from center to edge is 1.25cm, (i.e. a distance about 130X greater than the distance through the polyester) partially offsetting the higher permeability of gelatin. % Loss in ΔOD v Unlaminated Laminated 0 24hrs. 48hrs. 60hrs. 126hrs. Time in Desiccator at 50 o C Figure 44. Percent decrease in ΔOD v for a 3 Gy dose, following different times in a desiccator at 50 ºC.

122 Initial 24 hr 48 hr Absorbance (a.u.) Wavelength (nm) Figure 45. Spectral comparisons of absorbance of desiccated and normal unlaminated EBT film. Since the sensitivity of the films containing the lithium salt of PCDA as the active component change so dramatically when desiccated, we believe that water molecules are an integral part of the molecular structure. Further evidence for this can be seen in the changes in the spectral absorbance of the lithium PCDA polymer when irradiated film samples are desiccated. Figure 45 shows the changes in the absorbance spectra of exposed unlaminated film A after desiccation at 50 ºC. The changes in the absorption spectrum occur within 24 hours. The absorbance peaks do not shift on further desiccation thereafter. Figure 46 also illustrates this transition from hydrated to desiccated form at shorter time intervals. Seemingly, at 4 hours both forms exist, and after five hours at this temperature, the active component is desiccated. When the desiccated film is re-introduced to a humid environment, the sensitive layer absorbs water. However, instead of reverting to the spectrum prior to desiccation, a further shift to longer wavelength occurs (Figure 47). Since wavelength of absorbance depends on the structure of the backbone, we believe this shift signifies re-integration of water into the molecular structure. However, because the spectral absorbance differs from that prior to

123 Absorbance (a.u). 30 hours hours 24 hours 9 hours hours 7 hours hours 5 hours 4 hours hours hours 1 hour 0.2 unexposed Wavelength (nm) Figure 46. Absorbance of unlaminated EBT film after time in desiccator at 50 ºC (arbitrary dose, equivalent to about 10 Gy, delivered using UV light). 0.8 desiccated, 48 hours rehydrated, 96 hours Initial Absorbance (a.u.) Wavelength (nm) Figure 47. Spectral comparisons of absorbance of desiccated, rehydrated, and normal unlaminated EBT film irradiated to 3 Gy.

124 108 desiccation, we believe that the conformation and structure of the polydiactylene after rehydration differs from that prior to desiccation. It has also been found that the visible absorption spectra of the diacetylene polymers formed by radiation exposure of the two particle forms are different. The polymer from the hairlike form exhibits two peaks with the principal peak having an absorbance maximum at about 635nm. The spectra of polymer from the plate-like form is qualitatively similar, but the main peak is at approximately 670nm. Figure 48 shows absorbance spectra for the plate-like form of polymerized lithium PCDA superimposed on the spectrum of the polymerized hair-like form after desiccation and rehydration. This re-hydrated hair-like form has similar absorbance maxima as the unlaminated film B, that contains the plate-like arrangement of the active component. We take this to indicate that the structure of the polymerized diacetylene is the same in each. The data suggest that water is an important part of the structure of the monomers and polymers formed from the hair-like and plate-like of lithium pentacosa-10,12-diynoic acid. It is likely that water molecules are important in establishing the molecular structure of the hair- 0.8 "plate-like" form rehydrated "hair-like" form Absorbance (a.u.) Wavelength (nm) Figure 48. Absorbance spectra of exposed unlaminated films using plate-like form of polymer, and the rehydrated form of hair-like polymer.

125 109 like form and that in this form the spacing and orientation of the diacetylene units (LiPCDA) in adjacent diacetylene molecules is at least partly responsible for the high sensitivity observed for EBT. When the polymerized hair-like form loses water, it undergoes a re-configuration. The spectral shift suggests that the conjugated polymer backbone has undergone a twisting leading to a change in the overlap of the π-orbital system. When this desiccated polymer is then exposed to moisture, the water molecules are re-introduced in the polymer structure with further changes in the molecular configuration and the resulting structure that appears to be similar to the platelike form of the lithium PCDA. How this affects polymerization during further irradiation remains to be determined. IV. Conclusion Change in OD of GafChromic EBT film (measured over nm, corresponding to the main absorbance peak at 22ºC) for 1 Gy dose was shown to decrease by more than 10% when the temperature was increased from 22 ºC to 38 ºC. This decrease could not be attributed to a decrease in wavelength of maximum change in absorbance, λ max. A correction algorithm, using the shifting position of λ max with temperature and a temperature-dependent correction factor, was tested, but showed only moderate improvement over using a 10 nm range centered about λ max. The sensitivity of EBT to ionizing radiation was shown to also be a function of the hydration of the sensitive layer. Water influences the three-dimensional structure of the monomer crystals, and desiccating the samples shifted both the absorbance peak to a higher wavelength and decreased sensitivity. Rehydrating the samples created an alternate three-dimensional arrangement seemingly similar to that observed in EBT s predecessor, MD-55, along with a similar position in λ max and sensitivity to radiation. ACKNOWLEDGEMENTS The authors would like to thank Matt Filletti and Jason Smale of Princess Margaret Hospital for their work on the phantom, Dr. Kai Zhang of the Laboratory for Applied Biophotonics for fibercoupling the LED, and also Shelley Shih and Xiang Yu of the Advanced Materials Group of International Specialty Products. This work was in part funded by Research Studentship of The Terry Fox Foundation through an award (016646) from the National Cancer Institute of Canada, Scace Prostate Cancer Award, National Institutes of Health / National Institute on Aging (R21/R33 AG19381), and by the Fidani Center for Radiation Physics.

126 Maryanski M.J., Ibbott G.S., Eastman P., Schulz R.J., Gore J.C. Radiation therapy dosimetry using magnetic resonance imaging of polymer gels. 9Medical Physics 23: , Low D.A., Dempsey J.F., Venkatesan R., Mutic S., Markman J., Haacke E.M., Purdy J.A. Evaluation of polymer gels and MRI as a 3-D dosimeter for intensity modulated radiation therapy. Medical Physics 26: , Berg A., Ertl A., Moser E. High resolution polymer gel dosimetry by parameter selective MR-microimaging on a whole body scanner at 3 T. Medical Physics 28: , Wuu C.S., Schiff P., Maryanski M.J., Liu T., Borzillary S., Weinberger J. Dosimetry study of Re-188 liquid balloon for intravascular brachytherapy using polymer gel dosimeters and laser-beam optical CT scanner. Medical Physics 30: 132-7, Oldham M., Siewerdsen J.H., Kumar S., Wong J., Jaffray D.A. Optical-CT gel dosimetry I: Basic investigations. Medical Physics 30: , Chiu-Tsao S., Duckworth T.L., Patel N.S., Pisch J., Harrison L.B. Verification of Ir-192 near source dosimetry using GAFCHROMIC film. Medical Physics 31: 201-7, Hirata E.Y., Cunningham C., Micka J.A., Keller H., Kissick M.W., DeWerd L.A. Low dose fraction behavior of high sensitivity radiochromic film. Medical Physics 32: , Rink A., Vitkin I.A., Jaffray D. Suitability of radiochromic medium for real-time optical measurements of ionizing radiation dose. Medical Physics 32: , Rink A., Vitkin I.A., Jaffray D.A. Characterization and real-time optical measurements of the ionizing radiation dose response for a new radiochromic medium. Medical Physics 32: , Guillet J. Photopolymerization. In: Polymer Photophysics and Photochemistry: an introduction to the study of photoprocesses in marcomolecules. New York: Cambrdige University Press, p , 1985.

127 11. Baughman R.H. Solid-state synthesis of large polymer single crystals. Journal of Polymer Science: Polymer Physics Edition 12: , Baughman R.H., Yee K.C. Solid-state polymerization of linear and cyclic acetylenes. Journal of Polymer Science: Macromolecular Reviews 13: , Tsibouklis J., Pearson C., Song Y.P., Warren J., Petty M., Yarwood J., Petty M.C., Feast W.J. Pentacosa-10,12-diynoic Acid/Henicosa-2,4-diynylamine alternate-layer Langmuir- Blodgett films: synthesis, polymerization and electrical properties. Journal of Materials Chemistry 3: , Chu R.D.H., Van Dyk G., Lewis D.F., O'Hara K.P.J., Buckland B.W., Dinelle F. GafChromic dosimetry media: A new high dose, thin film routine dosimeter and dose mapping tool. Radiation Physics and Chemistry 35: , McLaughlin W.L., Yun-Dong Chen, Soares C.G., Miller A., Van Dyk G., Lewis D.F. Sensitometry of the response of a new radiochromic film dosimeter to gamma radiation and electron beams. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 302: , Saylor M.C., Tamargo T.T., McLaughlin W.L. A thin film recording medium for use in food irradiation. Radiation Physics and Chemistry 31: , Mack A., Mack G., Weltz D., Scheib S.G., Böttcher H.D., Seifert V. High precision film dosimetry with GafChromic films for quality assurance especially when using small fields. Medical Physics 30: , Klassen N.V., van der Zwan L., Cygler J. GafChromic MD-55: Investigated as a precision dosimeter. Medical Physics 24: , Rink A., Vitkin I.A., Jaffray D.A. Energy dependence (75 kvp to 18 MV) of radiochromic films assessed using a real-time optical dosimeter. Medical Physics 34: , Avena Bustillos R., Olsen C., Olsen D., Chiou B., Yee E., Bechtel P., McHugh T. Water vapor permeability of mammalian and fish gelatin films. Journal of Food Science 71: E202-E207, 2006.

128 CHAPTER 6: ENERGY DEPENDENCE OF GAFCHROMIC MD-55 AND EBT Portions of the following have been published as Energy Dependence (75 kvp to 18 MV) of Radiochromic Films Assessed Using a Real-time Optical Dosimeter by Alexandra Rink, I. Alex Vitkin, and David A. Jaffray in Medical Physics 34(2), p (2007) 112

129 113 I. Introduction Measurements of ionizing radiation dose are often required in diagnostic radiology, radiation therapy, health physics, and other scenarios. It is thus of interest to find a robust dosimeter that can be used across a wide range of radiation types and for various purposes, in order to decrease the complexity of radiation dosimetry and to permit assurance of appropriate radiation levels without excessive effort or cost. Such a dosimeter would have to meet several criteria, (1) including small size for in situ use, real-time response, and independence of measured signal across all x-ray energies of interest. Ideally, the detected dose from such a dosimeter should be equivalent to that delivered to water or tissue, so called energy-independent response. This will enable the dosimeter calibration at any beam energy for which the operator has confidence in the absolute dosimetry (e.g. 60 Co), and then its use with any beams without energy-dependence correction. Recently, two radiochromic films (GafChromic MD-55 and EBT) made by International Specialty Products (ISP, Wayne, NJ) have been considered as potential candidates for in situ real-time optical dosimetry. (1,2) Although GafChromic MD-55 (equivalent to MD referred to in some publications, (3) and henceforth referred to as MD-55) performed reasonably well with respect to ideal real-time dosimeter criteria, (1) traditional measurements (waiting some time after irradiation) of optical density showed a decrease in response per given dose as energy decreases. (4) The response of GafChromic HS (hereafter referred to as HS) relative to that of MD-55 for traditional measurements was recently summarized, (5) and appears to be similar across the energy range tested. Investigations of other versions of radiochromic films with the same sensitive material as currently used in MD-55 and HS showed the same trend. (6-8) On the other hand, GafChromic EBT (henceforth referred to as EBT) was suggested by its manufacturer to have a response to dose-to-water that is independent of energy since it has an effective atomic number (Z eff of 6.98 as quoted by the manufacturer) closer to that of water (Z eff = 7.3) than Z eff of 6.5 for MD-55. However, Z eff is just a first order approximation to how these complex films interact with photons at various energies. It gives no hint to whether all photon energies are equally effective at inducing polymerization within the sensitive medium. For this reason, it is important to investigate whether optical density of radiochromic films changes with photon beam energy. Several authors have recently reported on energy dependence of EBT. (9,10) One report used a very large range of dose rates for the different irradiations, (9) thus possibly introducing extraneous errors to the optical density measurements. The other reports

130 114 optical density measurements taken 24 hours after irradiation, making this data inappropriate for evaluation of energy independence of EBT in real-time dosimetry. Since the radiochromic media are being investigated for real-time dosimetry purposes, trends in the films response observed after some time has elapsed since irradiation need to be verified in real-time as well. In this paper, we compare the change in optical density as measured immediately at the end of irradiation for one Gy delivered to previously unexposed MD-55, HS, and EBT films for a number of beams of a range of photon energies (75 kvp to 18 MV). II. Methods and Materials Solid Water Phantom A 30 cm 30 cm 4 cm water equivalent phantom (Solid Water ) was made for the experiments (Figure 49). Two sets of inserts were designed: one for ion chamber measurements, the other for the radiochromic film measurements, where the latter is shown on the right of Figure 49. The former was made such that the center of the ion chamber (Model 2571, NE Technology Ltd., UK; calibration traceable to National Research Council) measuring volume Figure cm 30 cm 4 cm phantom with the film insert (assembled left, and without the top half on the right). The center of film positioned within the film holder is located in the middle of the 30 cm 30 cm plane. Delivery and collection optical fibers are positioned at 1.5 cm depth within the phantom, such that the interrogation light path is perpendicular to the film plane (as shown by straight white arrows). M.J. Butson (private communication, ).