Applied Radiation and Isotopes

Transcription

1 Applied Radiation and Isotopes 69 () 6 67 Contents lists available at ScienceDirect Applied Radiation and Isotopes journal homepage: Evaluation of dual g-ray imager with active collimator using various types of scintillators Wonho Lee a,n, Taewoong Lee a, Manhee Jeong b, Ho Kyung Kim c a Department of Radiologic Science, Korea University, Seoul 36-3, Republic of Korea b Department of Nuclear Engineering and Radiological Sciences, University of Michigan, Ann Arbor, USA c Department of Mechanical Engineering, Pusan National University, Pusan 69-73, Republic of Korea article info Article history: Received 7 February Received in revised form 3 May Accepted 6 July Available online 3 July Keywords: Dual imager Scintillator MURA Compton camera abstract The performance of a specialized dual g-ray imager using both mechanical and electronic collimation was evaluated by Monte Carlo simulation (MCNP). The dual imager consisted of an active collimator and a planar detector that were made from scintillators. The active collimator served not only as a coded aperture for mechanical collimation but also as a first detector for electronic collimation. Therefore, a single system contained both mechanical and electronic collimation. Various types of scintillators were tested and compared with each other in terms of their angular resolution, efficiency, and background noise. In general, a active collimator had the best mechanical collimation performance, and an LaCl 3 (Ce) active collimator provided the best electronic collimation performance. However, for low radiation energies, the mechanical collimation images made from both scintillators showed the same quality, and, for high radiation energies, electronic collimation images made from both scintillators also show similar quality. Therefore, if mechanical collimation is used to detect lowenergy radiation and electronic collimation is applied to reconstruct a high-energy source, either LaCl 3 (Ce) or would be appropriate for the active collimator of a dual g-ray imager. These results broaden the choice of scintillators for the active collimator of the dual g-ray imager, which makes it possible to consider other factors, such as machinability and cost, in making the imager. As a planar detector, showed better performance than other scintillators since its radiation detection efficiency was highest of all. & Elsevier Ltd. All rights reserved.. Introduction g-ray imaging has been applied to many fields, including industrial surveying, nuclear medicine, astronomical observations, and security. Reconstructing the radiation images requires collimation to filter or select incident g-rays. Hence, it is important to construct a collimator with high efficiency, fine resolution, and minimized noise. In most applications, mechanical collimation using a high-density material is employed to form an image in a position-sensitive detector (Fenimore and Cannon, 97; Guru et al., 996; Redus et al., 996). The radiation that passes through mechanical collimation such as a pinhole or multi-hole casts a shadowgram on the detector, and the shadowgram is converted to a radiation source image directly or after reconstruction. Other systems employ electronic collimation using sequential gamma ray interactions in a single detector or multiple detectors (Singh and Doria, 93a,b; Lehner, ). When n Corresponding author. Tel.: þ 9 6; fax: þ address: wonhol@korea.ac.kr (W. Lee). radiation undergoes Compton scattering in one position and is absorbed in the next position, the scattering angle can be calculated using the Compton formula with the deposited energy information, and a cone can be projected on a source plane based on the scattering angle and position information of the interactions. Mechanical collimation shows excellent performance for low-energy radiation (o kev), but the performance is not as good for high-energy ( kev) radiation because the stopping power of a mechanical collimator is inversely proportional to the incident radiation energy. By contrast, the performance of an electronic collimator improves with higher incident radiation energy because the energy uncertainty related to the angular uncertainty of a reconstructed image decreases as the incident radiation energy increases. Since mechanical and electronic collimation performance varies over the energy range, we have developed a new collimation method that combines the advantages of mechanical and electronic collimation. Several studies have developing approaches using both mechanical and electronic collimators. Gormley et al. (997) made the first attempt to use both mechanical and electronic collimation. In this system, a pinhole camera and a Compton 969-3/$ - see front matter & Elsevier Ltd. All rights reserved. doi:.6/j.apradiso..7.3

2 W. Lee et al. / Applied Radiation and Isotopes 69 () camera were combined in a separate module. Uritani et al. (997) added parallel plates to an electronically collimated camera to increase the signal-to-noise ratio. The parallel plates decreased the angular uncertainty, but they also decreased the field of view (FOV) and detection efficiency. In order to improve the detection efficiency and field of view, Smith et al. () replaced the parallel plates with a crude coded aperture. Lee et al. constructed an improved dual collimation gamma camera using LaCl 3 (Ce) scintillators (Lee and Wehe 7; Lee et al., 9). Since this system uses the information of both the Compton backprojections and the coded mask shadowgram itself, a single photon can contribute to both the mechanically and electronically collimated images, hence providing two pieces of information from a single photon. If the attenuation material (lead or tungsten) used for the mechanical collimation in dual g-ray collimation is replaced by detectors, the mechanical collimation not only attenuates the radiation for mechanical collimation but also constructs the electronic collimation using the sequential g-ray interactions in the collimation detector and the second detector. Since the mechanical collimator is active, both radiation passing through the mechanical collimator and scattered radiation can contribute to the collimation image; hence, the detection efficiency of a dual imager with an active mechanical collimator is inherently higher than that of previous systems. For our application, scintillators were chosen as the detector materials because of their high efficiency, ability to be used at room temperature, and feasibility in manufacturing. Since the various properties of a scintillator, such as its atomic number, density, and energy resolution, affect the efficiency and angular resolution of the reconstructed image, the image quality from dual collimation depends on the type of scintillator used. In this study, we simulated various scintillators to find the best ones as radiation detectors for a dual g-ray imager with an active collimator.. Material and methods As shown in Fig., our new dual g-ray imager consists of an active collimator and a planar detector. The scintillation materials are LaCl 3 (Ce), (Tl), (Na), and, whose densities are 3.79, 3.67,., and 7.3 g/cm 3, respectively. The electron densities for the scintillator materials calculated by density, atomic number, atomic weight and Avogadro s number are., 9.3 3,.3,. electron/cm 3, respectively. The distance between the active collimator and planar detector is cm, and a point mono-energetic radiation source is located cm from the active collimator. The total and pixel size of the image plane for mechanical collimation are 3 and cm, and the total and pixel angle of the image sphere for electronic Active Collimator (MURA) Planar Detector Source Fig.. Schematic diagram of the dual g-ray imager with active collimator. Fig.. Localization of a point source using Compton scattering.

3 6 W. Lee et al. / Applied Radiation and Isotopes 69 () 6 67 collimation are 36 (p field of view) and. Both collimator and detector are made from pixellated scintillators whose area is mm which is matched with the size of H Hamamastu photomultiplier tube. The area of each pixel is mm, which is close to the smallest size of scintillator available with current technology for LaCl 3 (Ce). The thickness of each pixel is mm, which is close to the maximum value for the finest pixels achievable by conventional manufacturing process for LaCl 3 (Ce). The geometry of the active collimator is based on a modified uniformly redundant array (MURA) (cf. Fig. ), which is one of coded apertures made by Gottesman and Fenimore (99). The MURA is a square array that is well matched with the shapes of the scintillation detectors and has an approximately % opening area. For the coded aperture, the recorded shadowgram on the planar detector is not similar to the source object because of many overlapping images that result from the encoding of the source distribution. Thus, to reconstruct the original object, the shadowgram must be decoded. The encoding process can be described as S ¼ðAnOÞþN where S is the image formed on the detector, A is the aperture, O is the source object, N is the noise function, and n indicates a correlation operator. In order to reconstruct the original source ðþ image, a correlation is applied as followed: O ¼ SnG ¼ðOnAÞnGþNnG¼OþNnG ðþ where G is a post-processing array that has been judiciously chosen so that AnG approximates a delta function. This reconstruction method is efficient, especially when the system is ideal and statistical noise is negligible. The aperture array A of MURA is constructed as follows (Gottesman and Fenimore, 99): if i ¼ >< A ¼fA ij g p i,j ¼, A if j ¼, ia ij ¼ ð3þ if C i C j ¼þ >: otherwise where p is odd prime number þ if i is a quadratic residue modulo p C i ¼ otherwise The decoding function G is constructed as follows: þ if iþj ¼ >< G ij ¼ þ if A ij ¼, ðiþjaþ >: if A ij ¼, ðiþjaþ ðþ ðþ Fig. 3. FWHM vs. incident radiation energy for various active collimators in mechanical collimation: (a) Planar detector: LaCl 3 (Ce), (b) planar detector: (Tl), (c) planar detector: (Na) and (d) planar detector:.

4 W. Lee et al. / Applied Radiation and Isotopes 69 () With the position and energy information obtained from the active collimator and the planar detector, an electronic collimation image can be reconstructed. As shown in Fig., if the first interaction position and energy (x, y, z, E ) deposited in the active collimator plus the second interaction position and photoelectric energy (x, y, z, E ) deposited in the planar detector are obtained, then the opening angle y for the Compton scattering is determined from Eq. (6) cosy e ¼ m ec E r E E s ¼ m ec E E ðe E Þ if E is known ¼ m ec E if E is unknown ðe þe ÞE and E ¼ E þe ð6þ If n is a unit vector along the line r and n is a unit vector along the line vector r cosy g ¼ n Un ð7þ Using Eqs. (6) and (7), one can calculate a conical shell of possible source positions containing (x, y, z ). We evaluated the performance of the scintillators over a broad range of energies ( 7 kev) of incident radiation and compared with each other. Since four different scintillators were used for both the active collimator and the planar detector, we studied sixteen combinations of detection systems. For electronic collimation, chance coincidence causing noise in the reconstructed image was also relatively calculated for each scintillator based on Eq. () (Knoll, ) r ch ¼ r s t r where r ch is chance coincidence rate, r s is random pulse rate and t r is resolving time of scintillator. The simulation code was MCNP and every photon was tracked using PTRAC card which is a built-in-function in the simulation code. The raw data from PTRAC were rearranged by Cþþ to select the related data for the dual collimation imaging. The simulation of Doppler broadening for each material was inherently included in the simulation code. In order to save the simulation time, the radiations from the source were initially directed to the area of the active collimator only. The total iteration number and the average running time for each simulation was 7 and s, respectively. The. GHz personal computer with GB RAM was used. 3. Results The g-ray images obtained by the dual collimation system were evaluated based on the angular resolution, efficiency, and background noise. The full width at half maximum (FWHM) and the peak value of the reconstructed point source were chosen to represent the angular resolution of the image and the efficiency, ðþ 6 x 6 6 x x 6 6 x Fig.. Peak counts vs. incident radiation energy for various active collimators in mechanical collimation: (a) Planar detector: LaCl 3 (Ce), (b) planar detector: (Tl), (c) planar detector: (Na) and (d) planar detector:.

5 6 W. Lee et al. / Applied Radiation and Isotopes 69 () 6 67 respectively. The background noise was calculated by measuring the relative standard deviation of counts recorded in peripheral pixels which were largely apart from a peak pixel. 3.. Mechanical collimation As shown in Fig. 3, the FWHM of mechanical collimation was almost constant regardless of the incident radiation energy: Dy mc ¼ tan ðc=dþ ð9þ where d is the distance between the active collimator and the planar detector and c is the width of the collimator elements. The calculated result based on Eq. (9) was.9, which was equivalent to the FWHMs obtained in simulation (cf. Fig. 3). The count of peak pixels decreased with incident radiation energy since the radiation energy was inversely proportional to the detection probability (cf. Fig. ). The peak pixel counts were highest when the active collimator material was, and this was particularly true when the energy of the incident radiation was high. The explanation for this behavior is that the atomic number and density are proportional to the radiation stopping power, and the radiation, which penetrated the active collimator and was detected in the planar detector, contributes to the background in shadowgram, resulting in a decrease in the peak pixel counts in the reconstructed image. The ratios of peak count obtained with vs. other active collimators were large for middle and high radiation energies (3 kev) but much decreased for low radiation energies because almost all radiation was attenuated at low energy regardless of the type of scintillator. For planar detector materials, also provided the highest efficiency, particularly above middle radiation energy; the reason was the same as for the active collimator. Background noise was caused by the statistical fluctuation in detected counts, the decreased contrast caused by radiation penetration through the active collimator, and the difference between the binary MURA matrix and real conditions. As shown in Fig., the relative standard deviation of the background pixels was highly dependent on the types of the active collimators at middle and high energies ( kev) since the penetration of radiation through the active collimator is inversely proportional to the atomic number and density of the scintillator. Above kev the relative standard deviation was also somewhat dependent on the type of planar detector because more radiation deposited full energy on the scintillators with the higher atomic number and density, which decreased the variance in pixel counts. For low-energy radiation (o kev), however, all the scintillators had relative standard deviations that were approximately equal. In summary, showed best performance for mechanical collimation when the incident radiation energy was above kev, while there was no large benefit from in detecting lower-energy radiation. Relative Standard Diviation Relative Standard Diviation Relative Standard Diviation Relative Standard Diviation Fig.. Relative standard deviation vs. incident radiation energy for various active collimators in mechanical collimation: (a) Planar detector: LaCl 3 (Ce), (b) planar detector: (Tl), (c) planar detector: (Na) and (d) planar detector:.

6 W. Lee et al. / Applied Radiation and Isotopes 69 () Electronic collimation The angular resolution of electronic collimation is determined by the energy, position, and Doppler broadening in the detectors. Since the energy resolution of a detector is finite, the measured energy inherently includes an energy uncertainty which contributes to an angular uncertainty in the Compton angle calculated by Eq. (6). The finite pixel size of the detectors causes the position broadening and Doppler broadening is related to the electrons being bound to the nucleus (Lee, 9). Since the pixel size is fixed for all scintillators and the Doppler broadening gives only small contributions to the angular resolution of the scintillators (Lee, 9), the energy resolution of the detectors is a main factor in determining the angular resolution of electronic collimation. As expected, the angular resolution of all the scintillators improved as the radiation energy increased (cf. Fig. 6) since the energy uncertainty is inversely proportional to the square root of the energy of the incident radiation. Due to its high energy resolution, LaCl 3 (Ce) showed the best performance among the active collimators, especially for incident radiation with energy below 7 kev, while the FWHM of was the largest since its energy resolution was the lowest among scintillators. However, for high energies ( MeV) there were only slight differences between the FWHMs of active collimators since the energy uncertainty was relatively small at high energies and thus the position uncertainty dominated the uncertainties causing angular broadening. The types of scintillator used for the planar detectors did not affect the FHWM values of the electronic collimation because we assumed that the incident radiation energy was known a priori in Eq. (6). The maximum pixel count increased initially with incident radiation, but, after a reaching saturation point, it decreased instead (cf. Fig. 7). This trend can be explained by the combination of the probability of Compton scattering in the active collimator and the detection efficiency in the planar detector. When the incident radiation energy increased, the probability of Compton scattering inversely proportional to the radiation energy, but not only was the detection efficiency related to the Compton scattering but also the photoelectric effect was inversely proportional to the incident radiation energy, so those two factors were competing with each other, resulting in a saturation point. The saturation point was also dependent on the type of the scintillator since the probability of radiation detection depended on the electron density of the scintillator. Among the tested scintillators, the LaCl 3 (Ce) active collimator provided the highest maximum counts at low energy (o kev), while provided the highest counts above middle energy ( kev). The types of planar scintillator had an effect on the counts of the peak Fig. 6. FWHM vs. incident radiation energy for various active collimators in electronic collimation: (a) Planar detector: LaCl 3 (Ce), (b) planar detector: (Tl), (c) planar detector: (Na) and (d) planar detector:.

7 66 W. Lee et al. / Applied Radiation and Isotopes 69 () 6 67 pixels because the detection efficiency was related to the electron density of the scintillator used in the planar detector. The planar detector provided the highest peak counts among scintillators. The background noise for electronic collimation was approximately ten times lower than that for mechanical collimation since for electronic collimation statistical fluctuation was the only cause of noise (cf. Fig. ). At low radiation energies, the reconstructed image from the LaCl 3 (Ce) active collimator showed the lowest noise level and the reconstructed image from with active collimator had the largest amount of noise. However, at high radiation energies the background noise in the active collimator was close to that obtained with LaCl 3 (Ce) or even less. As a planar detector, provided slightly less noise than other scintillators for the entire range of radiation energies. The relative chance coincidence rate was shown in Fig. 9 based on the decay time of each scintillator, which is linearly related with the resolving time of the detection system, and LaCl 3 (Ce) showed best performance. In summary, as an active collimator LaCl 3 (Ce) showed the best performance for electronic collimation when the incident radiation energy was less than MeV, while provided a large benefit in efficiency in detecting high-energy radiation ( kev). As a planar detector, was somewhat better than the other scintillators in terms of efficiency. The chance coincidence rate of LaCl 3 (Ce) was lowest among all scintillators.. Conclusion For mechanical collimation, a active collimator showed the best performance when the incident radiation energy was above kev, while there was no major benefit from in detecting lower-energy radiation. On the other hand, for electronic collimation, an LaCl 3 (Ce) active collimator showed best angular resolution and decent detection efficiency, especially below kev, while provided good angular resolution, the highest detection efficiency, and little background noise at high energies. Therefore, either LaCl 3 (Ce) or can be used as an active collimator for a broad range of radiation energies if mechanical and electronic collimation are applied in a dual mode. In other words, both scintillators were effective for use in mechanical collimation for low energies and in electronic collimation for high energies since their performances were close to each other in both energy ranges. (Tl) and (Na) active collimators provided performance that was equal to or less than the optimized scintillator in each of these condition. These results broaden the choice of scintillators for use as the active collimator in a dual g-ray imager, and hence we can consider other factors in manufacturing the imager, such as machinability and cost. As a planar detector, showed better performance than the other scintillators since its radiation detection efficiency was highest of all. Since the decay time of LaCl 3 (Ce) is at least time times shorter than those of other scintillators, LaCl 3 (Ce) could show relatively high performance for the measurement of high activity sources Fig. 7. Peak counts vs. incident radiation energy for various active collimators in electronic collimation: (a) Planar detector: LaCl 3 (Ce), (b) planar detector: (Tl), (c) planar detector: (Na) and (d) planar detector:.

8 W. Lee et al. / Applied Radiation and Isotopes 69 () Fig.. Relative standard deviation vs. incident radiation energy for various active collimators in electronic collimation: (a) Planar detector: LaCl 3 (Ce), (b) planar detector: (Tl), (c) planar detector: (Na) and (d) planar detector:. Relative Chance Coincidence Random Pulse Rate (cps) Fig. 9. Relative chance coincidence vs. random pulse rate for various active collimators in electronic collimation. the Ministry of Education, Science and Technology in Korea and a Korea University Grant funded by Korea University in Korea. References Fenimore, E.E., Cannon, T.M., 97. Appl. Opt. 7, 337. Guru, S.V., He, Z., Wehe, D.K., Knoll, G.F., Redus, R.H., Squillante, M.R., 996. Nucl. Instrum. Methods A 37, 6. Gormley, J.E., Rogers, W.L., Clinthorne, N.H., Wehe, D.K., Knoll, G., 997. Nucl. Instrum. Methods A 397,. Gottesman, S.R., Fenimore, E.E., 99. Appl. Opt., 3. Knoll, G.F.,. Radiation Detection and Measurement, third ed., p. 33. Lehner, C.E.,. Ph.D. Dissertation. University of Michigan. Lee, W., Wehe, D.K., 7. Nucl. Instrum. Methods A 79,. Lee, W., Wehe, D.K., Jeong, M., Barton, P., Berry, J., 9. IEEE Trans. Nucl. Sci 6 (3). Lee, W., 9. J. Korean Phys. Soc., 7. Redus, R., Squillante, M., Gordon, J.S., Bennett, P., Entine, G., 996. IEEE Trans. Nucl. Sci. 3 (7). Singh, M., Doria, D., 93a. Med. Phys.,. Singh, M., Doria, D., 93b. Med. Phys.,. Smith, L.E., Chen, C., Wehe, D.K., He, Z.,. Nucl. Instrum. Methods A 6, 76. Uritani, A., Clinthorne, N.H., Gormley, J.E., LeBlanc, J.W., 997. IEEE Trans. Nucl. Sci., 9. Acknowledgment This work was supported by a 3N Researcher Grant (- 377) funded by Korea Science and Engineering Foundation of