Signal Strength and SNR Enhancement Techniques for Frequency Domain Photoacoustic Radar Imaging

Transcription

1 Signal Strength and SNR Enhancement Techniques for Frequency Domain Photoacoustic Radar Imaging by Wei Wang A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Mechanical and Industrial Engineering University of Toronto Copyright by Wei Wang 2016

2 Signal Strength and SNR Enhancement Techniques for Frequency Domain Photoacoustic Radar Imaging Abstract Wei Wang Doctor of Philosophy Department of Mechanical and Industrial Engineering University of Toronto 2016 A method for improving photoacoustic imaging depth and signal-to-noise (SNR) ratio through an externally imposed thermal field has been developed. Uniform heating, microwave and ultrasound localized heating have been applied for PA image improvement. The microwave or ultrasound-induced elevated temperature was experimentally measured using thermocouples in ex-vivo bovine muscle. The measured temperature on the heated surface spot induced by microwaves and the temperature induced at the heated center of a HIFU (High Intensity Focused Ultrasound) focal area were compared with the theoretically estimated temperature. The analytical method showed excellent agreement with the measured results. The experimental results with uniform heating showed that the maximum imaging depth was increased by 20% by raising the temperature of absorbing biotissues uniformly from 37 to 43, and the SNR was increased by 8%. The microwave assisted localized heating can increase the imaging depth by 11%, and the SNR by 5%. The parameters making up the Gruneisen constant were also investigated. The studies showed that the Gruneisen constant of biotissues increases with temperature. The thermal expansion coefficient contributes to the increase of the Gruneisen ii

3 constant with temperature more than other relevant parameters (the specific heat capacity and the speed of sound) of ex-vivo bovine muscle. In addition to the thermal method, wavelength modulated differential PA radar is proposed for improving signal-to-noise ratio and early cancer diagnosis. This method is based on harmonic wavelength modulation which results in differential PA radar signal to enhance early cancer detection. Two chirp-waveforms out-of-phase modulating 680 nm and 800 nm laser beams can effectively suppress the background noise, greatly enhance the SNR, detect small variations in hemoglobin oxygen saturation levels, and be able to distinguish malignant tumors. The experiments were conducted at different hemoglobin oxygen saturation (StO2) levels of fresh sheep blood. Experimental results demonstrated the superior sensitivity by the differential PA method over other methods. iii

4 Acknowledgments I would like to express my sincere gratitude to my supervisor, Prof. Andreas Mandelis for his guidance, help, extreme patience, and generous support throughout my PhD study. I am truly thankful to his steadfast integrity and selfless dedication to my academic development. I would like to extend my appreciation to my committee members, Prof. Richard Cobbold and Prof. Anthony Sinclair for their valuable suggestions and comments on my research. They have routinely provided valuable insight into my research, and their advice has greatly contributed to the quality of the thesis. I would also like to extend my sincere gratitude to Dr. Kirk Michaelian and Prof. David Sinton for their valuable comments on the thesis. My appreciation extends to Prof. Markus Bussmann and Ms. Brenda Fung for their administrative assistance throughout my study. My gratitude is extended to Dr. Davoud Jahani for the specific heat capacity measurement by DSC. Appreciation is also expressed to Dr. Tse V. Chan from ECE for the microwave experimental setup and measurement. I would like to thank Dr. Bahman Lashkari for the initial Labview program of experimental setup, Dr. Xinin Guo for the theory of wavelength modulated experiment, Dr. Alexander Melnikov and Dr. Sreekumar Kaiplavil for assembling and testing the laser system, Mr. Sung soo Choi for the experimental setup of wavelength modulated experiment, and Ms. Edem Dovlo for the experimental setup of ultrasonic array imaging. I am grateful to my colleagues for their support and friendship: Dr. Sergey Telenkov, Dr. Mohammad Khosroshahi, Dr. Anna Matvienko, Dr. Nima Tabatabaei, Dr. Raymond Jeon, Dr. Koneswaran Sivagurunathan, Mr. Rudolf Alvi, Mr. Lilei Hu, Ms. Xian Li, and Mr. Ting Huan. Finally, I extend my gratitude to my mother and family for their endless love. This research would not be possible without their support. iv

5 Table of Contents Contents Acknowledgments... iv Table of Contents... v List of Abbreviations and Symbols... viii List of Tables... xv List of Figures... xvi Chapter 1 Introduction to Photoacoustic Radar Imaging Introduction Introduction to Photoacoustic Imaging for Early Cancer Detection Photoacoustic Imaging Method Research Motivation Research Objectives... 7 Chapter 2 Material and Instrumentation Preparation Material and Instrumentation CW Diode Laser Assembly Maximum Permissible Exposure Limit Introduction to FDPA Radar System Tissue Phantoms and Biological Samples Preparation Light Absorption Coefficient Measurement Summary Chapter 3 Theoretical Background of Frequency Domain Photoacoustic Theoretical Analysis of PA System Frequency Domain Photoacoustic Signal Generation v

6 3.2 Gruneisen Parameter Dependency with Temperature Thermal Field of Localized Heating Signal Processing Lock-in Method Matched Filter Compression Method Wavelength-modulated Differential Photoacoustic PA Signal of Differential Method Biomedical Application of Hypoxia Monitoring Summary Chapter 4 Temperature Effects on Photoacoustic Radar Module Thermally Enhanced Signal Strength and SNR Improvement of Photoacoustic Radar Module Using Heating to Increase the PA Signal Strength and SNR Experimental Validation of Thermally Assisted PA Summary Chapter 5 Localized Heating Enhancement of Photoacoustic Radar Signals for Tissue Diagnostic Imaging Localized Heating Assistance for the PA Radar System Introduction Localized Heating Assisted FDPA Radar by High Intensity Focused Ultrasound Comparison of the Heating Effects of HIFU and Microwave in Biological Tissues for PAR Measurement Localized Heating Assisted FDPA Radar by Microwave Summary Chapter 6 Frequency Domain Differential Photoacoustic Spectroscopy Frequency Domain Differential Photoacoustic Spectroscopy for Cancer Diagnosis Monitoring the Hemoglobin Oxygen Saturation Level for Cancer Diagnosis vi

7 6.2 Experimental Setup for the Differential Photoacoustic Measurement Results and Discussion Summary Chapter 7 Conclusions Thermally Enhanced Signal Strength and SNR Improvement of Photoacoustic Radar Module Localized Heating Enhancement of Photoacoustic Radar Signals for Tissue Diagnostic Imaging Wavelength Modulated Frequency Domain Differential Photoacoustic Radar Spectroscopy References vii

8 List of Abbreviations and Symbols Abbreviations AVE BPC CF CR CW DC DOS Deoxy EM EPR FDPA FFT FWHM GF HIFU Hb HbO2 ICG Description Average Black plastic color Contrast factor Change ratio Continuous wave Direct current Diffuse optical spectroscopy Deoxygenated Electromagnetic Electron Paramagnetic Resonance Frequency domain photoacoustic Fourier transform Full width at half maximum Green function High intensity focused ultrasound Hemoglobin oxygenated hemoglobin Indocyanine Green viii

9 IFFT Imag. IR LFM LOIS Inverse Fourier transform Imaginary Near infrared Linear frequency modulated Laser optoacoustic imaging system MPE The maximum permissible exposure (J/cm 2 ) MRI MW OCT Oxy PA PAR PTA PVCP RF SAR SNR SPION US Magnetic resonance imaging Microwave Optical coherence tomography Oxygenated Photoacoustic Photoacoustic radar Photo-thermo-acoustic Polyvinyl chloride-plastisol Radio frequency Specific absorption rate Signal to noise ratio Super paramagnetic iron oxide nanoparticles Ultrasound Roman based Symbol Unit Description ix

10 A Volt Amplitude of a reference signal r A Volt Amplitude of a detected signal s A Amplitude B Hz Frequency bandwidth b m Radial distance of a heating spot c m/s Speed of sound a C,C Forward or backward acoustic wave C A Correction factor C t Temperature constant C ox Oxy-Hb concentration C de Deoxy-Hb concentration C Hb Hb concentration coefficient C( x ) Variable c J/(kg C) Specific heat capacity p c J/(kg C) Specific heat capacity of blood b E J Hz/m 2 Energy fluence e ox Extinction coefficient of oxy-hb e de Extinction coefficient of deoxy-hb x

11 f g G Frequency Scattering anisotropy Green function G R Radial Green function G A Axial Green function h Complex conjugate of reference signal I, Ii W/cm 2 Intensity I0 Modified Bessel function of the first kind of order zero j Imaginary unit k Rad/m Acoustic angular wavenumber a k W/(m C) Thermal conductivity K second Chirp duration L m Axial distance m p Constant Pressure of signal Pr Power ratio q W/m 3 Volume heating source q W/m 2 Heat flux 0 r m Radial distance xi

12 r m Radial distance of absorbing medium R Cross-correlation output s r Reference signal s d Detected signal StO2 % Oxygen saturation S( x ) Variable t s Time T C Temperature Vr Reference signal Vs Volt Detected signal VLP Volt Output signal v m/s Velocity w % Mass fraction i w m 3 /(s kg) Blood flow rate b x Variable z m Axial direction (depth) z m Axial distance of absorbing medium * Z Complex conjugate operation Greek-Based Symbols xii

13 m 2 /s Thermal diffusivity C -1 Thermal expansion coefficient Gruneisen parameter Heat generation efficiency C Translated temperature µm Wavelength m -1 Light absorption coefficient a m -1 Light attenuation coefficient ef m -1 Light scattering coefficient s ' m -1 Reduced light scattering coefficient s s Np/m -1 Microwave or ultrasound amplitude absorption coefficient m ' Np/m -1 Microwave amplitude absorption coefficient m ' Np/m -1 Ultrasound amplitude absorption coefficient u Kg/m 3 Density Variable Standard deviation Acoustic velocity potential Phase signal r Phase of a reference signal xiii

14 s Phase of a detected signal rad/s Angular frequency r Angular frequency of reference signal s Angular frequency of detected signal Miscellaneous Symbols, Functions, and Abbreviations Variable * Complex conjugate operation rect(t) K 1, if t 2 0, otherwise xiv

15 List of Tables Table 2-1 Maximum permissible exposure calculation [49] Table 2-2 Number of samples for experiments Table 3-1 Thermal property equations for biotissue components (20 T 50 ) [59, 60].. 31 Table 4-1 The percentage difference comparison of water Gruneisen parameter elements Table 4-2 The percentage difference comparison of Gruneisen parameter elements of bovine muscle (75% water, 19% protein, 3% fat, 1% fiber, 1% carbohydrate, and 1% ash) Table 4-3 The percentage difference comparison of Gruneisen parameter elements for the second bovine muscle (81% water, 13% protein, 3% fat, 1% fiber, 1% carbohydrate, and 1% ash) Table 6-1 Comparison of StO2 concentration for tumor characterization study (DOS stands for diffuse optical spectroscopy) Table 6-2 Sensitivity comparison of non-invasive PA methods for StO2 measurement xv

16 List of Figures Figure 1.1. Photoacoustic phenomenon... 3 Figure 1.2. Photoacoustic tomography setup for non-invasive imaging of rat brain in vivo with skin and skull intact (modified version) [27]... 5 Figure 2.1 The fiber coupled laser diode and fast current modulator... 9 Figure 2.2 CW laser system Figure 2.3 Measured output power of 800 nm laser system dependence on the input signal amplitude peak voltage (V) Figure 2.4 Measured output power of 680 nm laser system dependence on the input signal amplitude peak voltage (mv) Figure 2.5 Experimental setup for FDPA measurement Figure 2.6 Signal processing schematic in the frequency domain Figure 2.7 The measured amplitude of PA signal Figure 2.8 Simulated single chirp excited PA phase signal (phase amplitude in radians) Figure 2.9 The result of Phase measurement Figure 2.10 Relation between light absorption coefficient a and BPC concentration in PVCP phantom [50] Figure 2.11 Experimental setup for measuring light absorption coefficient with temperature Figure 3.1. PA source geometry with light absorption Figure 3.2 Amplitude of PA signal as a function of distance (depth in the absorber) Figure 3.3 Amplitude of PA signal as a function of frequency xvi

17 Figure 3.4 Phase of PA signal as a function of distance (depth in the absorber) Figure 3.5 Phase of PA signal as a function of frequency Figure 3.6 Amplitude of PA signal as a function of light absorption coefficient Figure 3.7 Phase of PA signal as a function of light absorption coefficient Figure 3.8. Thermal expansion coefficient of water dependence on temperature [52] Figure 3.9. Specific heat capacity of water dependence on temperature [53] Figure Dependence of the speed of sound in water on temperature [54] Figure Gruneisen parameter of water dependence on temperature Figure 3.12 Estimated thermal expansion coefficient of bovine muscle as a function of temperature, (a) bovine muscle composition as 75% water, 19% protein, 3% fat, 1% fiber, 1% carbohydrate, and 1% ash; (b) bovine muscle composition as 81% water, 13% protein, 3% fat, 1% fiber, 1% carbohydrate, and 1% ash Figure 3.13 Estimated specific heat capacity of bovine muscle as a function of temperature, (a) and (b) have the same meaning as in the previous figure Figure 3.14 Experimental setup for speed of sound measurement Figure 3.15 Dependence of speed of sound of water on temperature Figure 3.16 Dependence of speed of sound in bovine muscle on temperature Figure 3.17 Estimated temperature dependence of the Gruneisen parameter of bovine muscle, (a) and (b) carry the same meaning as in Figures 3.12 and Figure Geometry of MW or US heating source Figure Diagram of the lock-in detection method Figure 3.20 Linear frequency modulated input chirp waveform (up chirp) xvii

18 Figure 3.21 Matched filter response chirp waveform (down chirp) Figure 3.22 Output results of matched filter compression method Figure 3.23 Absorption spectra of hemoglobin expressed as extinction coefficient of oxyhemoglobin eox and deoxy-hemoglobin ede [72] Figure 3.24 PA differential amplitude dependence on two lasers power ratio Figure 4.1 The percentage difference of Gruneisen parameter elements of water from 20 to Figure 4.2 The percentage difference from 20 to 45 of Gruneisen parameter elements of bovine muscle comparison Figure 4.3 The percentage difference from 20 to 45 of Gruneisen parameter elements for second bovine muscle (81% water, 13% protein, 3% fat, 1% fiber, 1% carbohydrate, and 1% ash) Figure 4.4 PA Signal dependence on Temperature (Measured on ink solution, light attenuation coefficient μeff=3.1 cm -1 ) Figure 4.5 PA Signal dependence on temperature (Measured on ex-vivo bovine muscle, averaged light attenuation coefficient μeff=5.9 cm -1 ) compared to estimated results (bovine muscle composition) Figure 4.6 (a) Light attenuation coefficient dependence on temperature; (b) The phase of the PA radar signal dependence on temperature Figure 4.7 (a) Experimental setup for PA radar imaging depth dependence on temperature; (b) PA radar imaging depth dependence on temperature (measured on ex-vivo bovine muscles) Figure 4.8 Experimental setup for PA radar imaging with transducer array Figure 4.9 Image generated by PA signals at 32 C, CF = Figure 4.10 Image generated by PA signals at 38 C, CF= xviii

19 Figure 5.1 Block diagram of the HIFU field pressure measurement Figure 5.2 Experimental setup for HIFU heating Figure 5.3 (a) Normalized pressure distribution of HIFU field in axial direction (1.1 MHz); (b) Normalized pressure distribution of HIFU field in radial direction (1.1 MHz) Figure 5.4 Experimental setup of HIFU assisted FDPA system Figure 5.5 Comparison of uniform heating and HIFU heating effects on FDPA signals Figure 5.6 (a) Measured temperature on a heated surface spot of bovine muscle vs. time compared with the analytically estimated MW-heating temperature; (b) Measured temperature at the center of the HIFU focal area vs. time compared with the theory Figure 5.7 Experimental setup for microwave heating Figure 5.8 Geometry of microwave antenna Figure 5.9 Experimental setup of the microwave coupled PAR system Figure 5.10 Measured PAR cross-correlation amplitude peak from ex-vivo bovine muscle with microwave and uniform heating Figure 5.11 (a) Experimental setup; (b) PAR signal imaging depth study under microwave heating Figure 5.12 (a) Picture of a 4 mm wide ex-vivo bovine muscle sample; (b) 1-D image at 37 with optical scattering (CF=1.9); (c) Same imaging conditions as in (b) with added microwave heating (CF=2.1) Figure 6.1 Experimental setup for wavelength modulated differential FDPA radar system Figure 6.2 The amplitude of single laser (680nm or 800nm) excited photoacoustic signals dependence on StO2 Concentration Figure 6.3 Differential amplitude of PAR signal with increasing P r xix

20 Figure 6.4 Differential PAR amplitude dependence on power ratio P r Figure 6.5 Sensitivity comparison of single-laser (680 and 800nm) excited PA amplitude CR and differential PA amplitude CR at power ratio ( P r = 0.85) dependence with StO2 Concentration.. 88 Figure 6.6 Sensitivity comparison of single laser (680 and 800nm) excited PA SNR change ratio (CR) with differential excited PA ( P r = 0.85) SNR-CR dependence with StO2 Concentration xx

21 1 Chapter 1 Introduction to Photoacoustic Radar Imaging In this chapter, cancer screening methods, the photoacoustic phenomenon, and the advantages of photoacoustic imaging system for early cancer detection are discussed. The research motivation and objectives are presented. In the end, the structure of thesis is outlined. 1 Introduction 1.1 Introduction to Photoacoustic Imaging for Early Cancer Detection The human body is made up of many different cell types. Cancer occurs when some cells grow and divide to produce other cells in a disorderly manner and new cells form a tumor when the body does not need them. The oldest description of cancer dates back to about 3000 B.C. in Egypt [1]. Hippocrates of Cos ( B.C.) named cancer as karkinoma (carcinoma) because a tumor looks like a crab where it has a main-core body and leg-like spreading parts [2]. Hippocrates is also credited for distinguishing the differences between benign tumors and malignant tumors [2]. Breast cancer is the leading cause of cancer-related death among women and remains the most common malignancy worldwide [3]. Early cancer detection can reduce the rate of death when followed by treatment and the detected tumors can only be effectively treated before metastasize (spreading to other organs or parts) [4]. X-ray based tomography is a good combination of sensitivity and specificity, and it is the most popular cancer screening tool [5]. The sensitivity and specificity of X-ray mammography are affected by tissue density, which in turn is influenced by age, body mass index, genetic tendency, etc. [5]. The results can also be affected by the technical execution of the scan and the methodology employed [6]. The X-ray mammography can cause discomfort to some patients. The examination process of mammography also exposes the tissue to ionizing radiation, which could cause radiation carcinogenesis [7]. Ultrasound (US) imaging is also known as sonomammography, which is often used as a follow-up test for an abnormal tissue examination result [7]. US is non-invasive, inexpensive, convenient, and not harmful to human tissues. US can distinguish different tissue layers (difference of mechanical property) within the tissues [8]. Despite these advantages, US specificity is limited (acoustic characteristics overlap between

22 2 benign and malignant tumors) and it cannot distinguish either oxygen saturation or the concentration of hemoglobin. Magnetic resonance imaging (MRI) uses radio waves and magnetic fields to change the alignment of hydrogen nuclei, and the change response creates the images [7]. MRI has demonstrated excellent results with respect to the sensitivity of cancer detection [7, 9]. MRI screening is not a perfect tool and there are still many uncertainties in the clinical practice [10]. Although in general, MRI is more sensitive to cancer detection than mammography or ultrasound, it can also miss some cancers. It is often recommended to be used in combination with mammography or ultrasound to avoid the false positive detection rate [10]. In addition, MRI screening requires a contract agent for signal enhancement, and it is more expensive than mammography and ultrasound [11, 12]. Intense efforts are under way to overcome the limitations of X-ray mammography, US, and MRI for early cancer detection. Researchers are also looking for alternative imaging methods. In recent years, photoacoustic imaging has been introduced for early cancer detection. This technique combines the advantages of high contrast due to light absorption with high resolution and penetration depth of the ultrasound signal. Photoacoustic (PA) or photo-thermo-acoustic (PTA) tomography are imaging techniques based on the photoacoustic effect. The photoacoustic effect was first discovered by Alexander Graham Bell in In his experiments, he found that acoustic signals can be generated by illuminating objects with chopped sunlight [13]. In PTA tomography, the object is irradiated by pulsed or modulated continuous wave laser beams. Some light is absorbed and partially converted into heat that is then converted into increased pressure and thermoelastic expansion. Finally, changing pressure generates ultrasonic waves [14]. The PA technique is based on the detection of acoustic signals generated by the light absorption of tissue, which combines the advantages of acoustic and optical methods. Research on PA imaging made very little progress after its discovery because of a lack of suitable sources for generating acoustic waves. In the 1970s, with the emergence of lasers, researchers regained interest in the PA phenomenon [15]. Kreuzer reported PA effects on detecting gas constituents using a laser as a light source in 1971 [15]. Since then, much PA-related research has been conducted. Nowadays, PA technology is widely used in biology, engineering, physics, and medicine [16, 17]. The principle of PA can be described as the absorption of electromagnetic (EM) radiation energy

23 3 that is then converted into heat in the medium thereby producing thermal expansion. If the heat source changes continuously, thermal expansion changes correspondingly and then acoustic waves are generated in the medium [13] (as shown in Figure 1.1). Generally speaking, two kinds of EM sources are used to generate PA signals: pulsed or modulated continuous waves (CW). Figure 1.1. Photoacoustic phenomenon With a properly controlled illuminating source, the PA technology can be used for biomedical imaging. It has been demonstrated that optical and RF wave absorption are highly related to the molecular formation and constitution of biological tissues. Therefore, PA signals can contain information regarding the functional and molecular nature of the tissues [14]. Although the pure optical detection method can be used to detect optical absorption, the detection result is limited by reflection of light, and is therefore less sensitive than ultrasound detectors [14]. The optical method can only be used for small depth detection: up to 1 cm, but with poor resolution [18]. Near infrared (IR) wavelengths exhibit low light absorption coefficients and relatively low scattering in cross-sections of biotissues, so the near IR region ( nm) allows light to penetrate deeper into the tissue, up to several centimeters [18, 19]. The maximum depth ultrasound imaging can reach is up to 15 cm with good resolution [20]. PA tomography has the advantage of combining optical tomography and ultrasonic tomography. PA imaging can offer optical contrast sensitivity like diffuse optical tomography and also can offer high spatial resolution similar to ultrasonic imaging. The optical absorption of biological tissue is related to its molecular composition and varies from tissue to tissue. Optical absorbers in tissues are oxygenated hemoglobin (HbO2), deoxygenated hemoglobin (Hb), water, and lipids [18, 21]. The molecular concentration of the oxygenated hemoglobin and deoxygenated hemoglobin has been used to identify malignant tumors [22]. Myoglobin and hemoglobin are the main optical absorbers in muscle tissues and the

24 4 concentration of myoglobin is much higher than hemoglobin. Studies show that myoglobin content is 0.41 mmol/kg and hemoglobin content is only mmol/kg in bovine muscle [23]. Myoglobin and hemoglobin are present in oxygenated or deoxygenated forms in ex-vivo tissues but deoxygenated concentrations are higher. At the surface of ex-vivo tissues, the degree of oxygenated concentrations is higher. The first PA system for breast cancer study was developed by Esenaliev et.al. [24]. The system was named laser optoacoustic imaging system (LOIS), and used a 1064 nm Q-switched pulsed laser source and an array ultrasound transducer as detector. The system detected a 2 mm absorber at the depth of 60 mm within a 100 mm thick breast phantom [24]. 1.2 Photoacoustic Imaging Method PA imaging can be classified by the excitation source, imaging construction method, or the applications for which it is used. Generally speaking, two kinds of laser sources can be used to generate PA signals: pulses and modulated continuous laser sources. The pulsed laser is the most popular source for PA research. The pulsed laser excitation PA system has the advantages of strong signals and easier absorber depth determination. However, it also has many disadvantages. One of the drawbacks of the pulsed laser PA system is the bulky and relatively expensive Q-switched nanosecond system. Such systems are difficult to operate, and the maintenance cost is very high. The pulsed PA system requires wide-band ultrasound detectors, which can increase system noise [25]. The pulsed laser generates inherently bipolar shaped signals that can have adverse effects on the spatial resolution of imaging [26]. A PA system has been used for rat brain imaging [27]. The system setup for rat brain imaging is shown in Figure 1.2. This system adopted a 532 nm pulsed laser to generate shortened laser pulses and a wide band with center frequency at 3.5 MHz ultrasound transducer was used for PA signal detection. The rat brain structure was accurately mapped noninvasively [27].

25 5 Figure 1.2. Photoacoustic tomography setup for non-invasive imaging of rat brain in vivo with skin and skull intact (modified version) [27] The recently developed intensity-modulated near-ir continuous wave (CW) laser source is a new alternative approach to PA imaging [28]. The CW laser source (laser diode) is inexpensive, compact, and portable. In addition, the narrow bandwidth detection signals with high duty cycles can significantly reduce the incoherent noise of the measurement. As such, the CW mode PA system is a promising technique for clinical applications. Lashkari and Mandelis have conducted a comprehensive comparison of a pulsed PA system with a CW PA system [29, 30]. It was found that the SNR, resolution and contrast between pulsed PA and CW PA are small in practice. However, the performance of the CW PA system can be significantly improved through signal processing. The CW PA system has more advantages in signal processing than pulsed PA. The experimental results and theoretical analysis demonstrated that CW PA is a competitive and alternative method to the conventional pulsed PA in terms of key imaging parameters: SNR, resolution, and contrast [29, 30].

26 6 1.3 Research Motivation The limitations of PA imaging for clinical application have been investigated. The performance characteristics of PA imaging systems include the maximum imaging depth, sensitivity, resolution, and contrast [25, 27]. The PA imaging system has good sensitivity, resolution and contrast, but the major limitation is the penetration of light [25-27]. Light intensity decreases exponentially with penetration depth because of strong scattering and absorption in biological tissues [18, 19]. The optical window in the 700 to 900 nm range allows light to penetrate relatively deeper [18, 19]. In addition, system parameter optimization, sophisticated signal processing, and higher laser power can also improve image quality [30]. Moreover, the PA signal can be enhanced by contrast agents, such as Indocyanine Green (ICG) dye [31], and Silicacoated super paramagnetic iron oxide nanoparticles (SPION) [32]. Even though these methods can improve SNR and imaging depth significantly, contrast agents are not without health risks and are, therefore, invasive. The widely accepted body temperature for a healthy adult is approximately 37 C [33, 34]. Studies show that raising the temperature of the whole human body up to 41.8 C for less than an hour is generally safe [35]. Studies also indicate that most normal tissues such as liver, kidney, and muscles can withstand temperatures up to 44 C for 30 minutes [36-38]. A relationship between thermal exposure time and temperature in most normal tissues is described by [38]: t t C 2 1 ( T1 T2 ) t (1-1) where t 1 and t2 are the exposure times at temperature T 1 and T 2, respectively. C t is 0.5 for temperatures above 43 C, and 0.25 for temperatures below 43 C [38]. The 43 C boundary is arbitrarily chosen from the available data, and is related to the normal physiological temperature of tissues [37-39]. According to Eq. (1), the heating effects of 60 minute exposure time at 43 C are equal to 30 minutes at 44 C, and to 15 minutes at 45 C. The results show that the maximum safe exposure time is shorter at higher temperatures. Many studies agree that the estimated exposure time results are generally safe for most normal tissues [36, 37]. However, central nervous system tissue is the most sensitive to temperature and damage was found after exposure to temperature over 42 C for longer than 30 minutes [40]. The skin is less sensitive to increased temperature than other tissues and the recommended maximum contact temperature without burn

27 7 injuries is 43 C with a maximum time of 8 hours [41]. The safe exposure time of skin is shortened by about 50% for each temperature degree increase up to 50 C, and it sustains no injury when exposed to 60 C for only a few seconds [41]. Temperature can affect PA signals: the PA signal is stronger at higher temperatures [42, 43]. Therefore, imaging quality and depth can be improved by heating methods. However, a detailed study of the PA signal increase that occurs with temperature has not been reported to the best of our knowledge. Differential PA measurements using two modulated lasers each emitting at a different wavelength can efficiently suppress background noise and improve the signal-to-noise ratio of the measurement significantly. Tissue oxygen saturation (StO2) was reported for peripheral arterial disease assessment [44]. Tumor hypoxia is associated with tumor propagation and malignant progression [45]. The monitoring of hemoglobin oxygenation level in real time is an important method for cancer diagnosis [45]. 1.4 Research Objectives In this research, we investigate the details of the temperature dependent parameters influencing the PA signal and we present techniques (uniform heating and localized heating) for using heat to increase the PA signal-to-noise ratio and improve the PA imaging depth. Also studied are the PA signal dependence on temperature, specifically of the elements making up the Gruneisen parameter, and the resulting PA imaging depth improvement at increased temperatures. Furthermore, a novel noninvasive wavelength multi-frequency modulated differential photoacoustic radar spectroscopy for early cancer diagnosis is investigated with a goal to improve the measurement sensitivity for early cancer detection. Thesis outline: Chapter 1 introduces the PA imaging, research motivation, and research objectives. Chapter 2 discusses the experimental instrumentation, material preparation, and the safety concerns of laser radiation.

28 8 Chapter 3 describes the theoretical background of signal generation of frequency domain photoacoustic (FDPA), temperature dependence PA signals, localized thermal field, signal processing in the frequency domain, and wavelength modulated differential photoacoustic radar spectroscopy. Chapter 4 presents a thermal method (uniform heating) for enhancing signal strength and SNR improvement of photoacoustic radar module. Chapter 5 proposes a localized heating method for enhancing photoacoustic radar signals for tissue diagnostic imaging. Chapter 6 introduces a wavelength-modulated differential frequency domain photoacoustic radar method for early cancer diagnosis. Chapter 7 gives conclusions of the dissertation.

29 9 Chapter 2 Material and Instrumentation Preparation The material and instrumentation of a frequency domain photoacoustic radar system, and the maximum permissible exposure limit of laser exposure for skin are described in this chapter. 2 Material and Instrumentation 2.1 CW Diode Laser Assembly The FDPA system requires a CW laser to be modulated at frequencies up to 5 MHz. As such, a passively cooled, fiber coupled diode laser for medical application (JOLD-7-BAFCM-12, Jenoptik) was selected for the experiments (as shown in Figure 2.1) [46]. The diode is driven by a fast current diode modulator FM 15 (MESSTEC, Germany), which can drive arbitrary current waveforms into a laser diode in very short rise (30 ns) and fall times (25 ns) [47]. Figure 2.1 The fiber coupled laser diode and fast current modulator The strong connection (shown in Figure 2.1) between the diode laser and modulator is essential for the high frequency modulation. Caution must be taken when connecting the diode laser to the current modulator (driver) by electronic soldering. The maximum permissible soldering temperature is 210 C for less than 10 seconds [46]. The soldered connection between the current modulator and diode laser should be a strong joint to ensure high current flow: up to 10 A in very

30 10 short time (ns) [46] [47]. An error in the connection may cause severe damage to the laser diode, driver, or other components. A static free environment is mandatory for the soldering connection operation [47]. The laser device has to be operated between 15 C to 30 C [46]. In the CW mode, the laser diode and driver will generate a large amount of heat, and the heat has to be removed by forced air cooling (cooling fans). A 3M TM thermal interface pad (5519, Digikey) with high thermal conductivity was used for mounting the laser module on the heat sink for efficient heat conduction. A TEC controller (Arroyo 5305; Arroyo Instruments) was used for the diode temperature control. The power supply for the driver was a 5V output AC-DC converter ( ND, Digikey) shown in Figure 2.2 [48]. All components were assembled into a metal box as shown in Figure 2.2. Switches were used to control the power or other operations as shown in the Figure. Three switches were employed to control the power sources. The key lock switch was used to lock the power supply of the whole system. One pushbutton power switch was employed to control the cooling power, and another pushbutton switch was employed to control the power for the laser diode and driver. The cooling fans should be started earlier in the startup process. For the shutdown process, the cooling fans should be turned off later. Two LED lights were used as indicators. The modulating signals were input into the system through a BNC socket and can be monitored through another socket in the front panel. The testing results showed the modulating frequencies can reach up to 5 MHz, meeting the requirement for this project. The output power of the 800 nm and 680 nm laser systems was measured by an optical power meter (1918-C, Newport). The measurement was repeated 5 times. The measured output power of the 800 nm and 680 nm systems is shown in Figure 2.3 and 2.4, respectively. From the figures, we can see the output power increases linearly with the peak voltage of the input modulating signals. The maximum output power of the 800 nm laser can reach up to 3.5 W with 1 MHz sinusoidal signal modulation, and the maximum output power the 680 nm laser can reach is up to 1.8 W with 1 MHz sinusoidal signal modulation.

31 11 Figure 2.2 CW laser system Figure 2.3 Measured output power of 800 nm laser system dependence on the input signal amplitude peak voltage (V)

32 12 Figure 2.4 Measured output power of 680 nm laser system dependence on the input signal amplitude peak voltage (mv) 2.2 Maximum Permissible Exposure Limit Laser light can burn the retina of eyes, skin, and cause other permanent injuries. Biological damage to eyes or skin is negligible under the MPE of radiation exposure. Therefore, it is necessary to minimize the risk of laser exposure damage. MPE (maximum permissible exposure) is an important parameter for the safety control, which is defined as the highest laser radiation power or energy density (J/cm 2 ) that a person may be exposed to without hazardous effects or biological changes [49]. The safety standard from ANSI [49] gives the definition of the exposure time under the laser exposure as a function of wavelength, pulse duration time, exposure aperture, and time. The MPE is calculated based on the laser wavelength and exposure time. Eyes can be protected by wearing certified laser safety glasses (Thorlabs). Table 2-1 shows the calculation of MPE for skin exposure. In the table, the MPE is different under different exposure times: 10-9 t <10-7, or 10-7 t <10. CA is a correlation factor, which is 1 for 680 nm laser [49]. For 800 nm laser, the factor is computed by [49]: 2( 0.7) C A 10 (2-1)

33 13 where is the wavelength (µm). For the 800 nm laser, CA is: 2(0.80.7) C A (2-2) Table 2-1 Maximum permissible exposure calculation [49] Wavelength (µm) Exposure time (s) MPE (J/cm²) 0.4 to to CA to CA t 1/4 For 400 continuous 1 ms long chirps with the total exposure time 0.4 s, the MPEs for 800 and 680 nm laser radiation are calculated as: MPE C J / cm 1/4 2 s nm A (2-3) MPE C J / cm 1/4 2 s nm A (2-4) For a single chirp with 1 ms exposure time, the MPE is: MPE C J / cm 1/4 2 s nm A (2-5) MPE C J / cm 1/4 2 s nm A (2-6) The laser exposure time should be lower than the recommended MPE, not only for the single chirp exposure but also for chirp-train exposure. 2.3 Introduction to FDPA Radar System A FDPA radar system diagram is shown in Figure 2.5. A continuous wave (CW) diode laser emitting at 800 nm or 680 nm was used for the experiments. The laser beam size was 3.5 mm. Linear frequency modulated (LFM) chirp signals (0.3 MHz to 1.3 MHz, 1 ms long) were utilized for laser modulation. The chirp signals were generated by LabView software, uploaded to the digital card NI PXI-5421, and synchronized with the data-acquisition process. A focused ultrasound transducer (Panametrics-NDT, V314 with -6dB range from 0.59 to 1.2 MHz, focal distance 1.9 inches) was employed as a photoacoustic signal detector. The detected signals were

34 14 amplified by a pre-amplifier (5676, Parametrics-NDT) first and then were sent to a digital dataacquisition card NI PXIe-5122 (National Instrument, Austin, Texas). A homemade heater was used to heat the water in a water-tank. Two thermocouples (K type, Omega) were employed for temperature measurements. One was used to measure the temperature of distilled water, and the other one was used to measure the temperature of the tested sample immersed in the water. The matched filter compression method was used for signal processing. The signal processing flowchart is shown in Figure 2.6. First, a reference signal sr ( t ) was processed by Fourier Transformation, and followed by the complex conjugate operation. Then the result was computed with the Fourier Transform (FFT) result of the detected signal s ( t ). The result was transformed by an Inverse Fourier Transform (IFFT) thereafter and the final results of the matched filter output s amplitude and phase were obtained. The PA measurement results are shown in Figures 2.7 and 2.8. The amplitude of the PA result is shown in Figure 2.7 and the phase is shown in Figure 2.8. Figure 2.8 shows a simulated PA phase generated by a single chirp. From this figure, it can be seen that the phase changes sharply from 0 to -39 radians at the corresponding delay time (the multi-phase PA signal was generated). A detailed analysis of PA phase with frequency is described in Chapter 3.1. It was found after averaging the phase results, the inverse of the standard deviation of the phase signals generated a peak as shown in Figure 2.9 [30]. d

35 15 Figure 2.5 Experimental setup for FDPA measurement Figure 2.6 Signal processing schematic in the frequency domain

36 16 Figure 2.7 The measured amplitude of PA signal Figure 2.8 Simulated single chirp excited PA phase signal (phase amplitude in radians)

37 17 Figure 2.9 The result of Phase measurement 2.4 Tissue Phantoms and Biological Samples Preparation Tissue phantoms and biological samples were used in the experiments to simulate the optical absorption of human tissues. Plastisol is made of polyvinyl chloride-plastisol (PVCP) with a certain amount of black plastic color (BPC) to resemble the absorption properties of the tissue. PVCP (M-F, USA) is a non-toxic plastic material and white opaque solution. It does not absorb light and becomes translucent after heating to a high temperature (about 200 C) for a certain time. The BPC (M-F, USA) is made of CI Pigment Black 7, which is added into PVCP as absorbing material. Figure 2.10 shows the relationship of the light absorption coefficient with BPC concentration in a PVCP phantom. The BPC amount should be carefully chosen according to the required absorption coefficient. After the PVCP phantom is synthesized, it becomes insoluble in water. PVCP phantoms have been widely used for optoacoustic, ultrasound, and biomedical applications [50].

38 18 Figure 2.10 Relation between light absorption coefficient a and BPC concentration in PVCP phantom [50] Samples of ink solution and ex-vivo bovine muscle were also prepared for the experiments. The ink solution was prepared by adding a small amount (about 1%) of liquid ink (Lamp black, Cotman) into distilled water (7 samples were prepared). The measured light attenuation coefficient of the ink solution at 20 is 3.1 cm -1 (mean) and with a standard deviation of 0.07 cm -1. The 800 nm laser was used for the attenuation coefficient measurement. The ex-vivo bovine muscle was purchased from a local market, and the samples were packed and kept at 4 in a lab refrigerator. The measured light attenuation coefficient of ex-vivo bovine muscle at 800 nm was 5.9 cm -1 (mean) with a standard deviation of 0.11 cm -1. The experimental tissue samples were cut from the stored bovine muscle and wrapped with a very thin transparent plastic film for the experiments. Nine samples were used for the attenuation coefficient measurements and the measurements were repeated 3 to 5 times. A list of the number of samples for all experiments is shown in Table 2-2.

39 19 Table 2-2 Number of samples for experiments Experiment description Number of samples Light attenuation coefficient of bovine muscle with temperature 9 Light attenuation coefficient of ink solution with temperature (ink solution) 7 Speed of sound of water (distilled water) 5 Speed of sound of bovine muscle 4 PA Signal dependence on Temperature (ink solution) 7 PA Signal dependence on Temperature (bovine muscle) 8 Imaging depth by uniform heating study (bovine muscle) 5 Temperature measurement with HIFU heating (bovine muscle) 7 Temperature measurement with Microwave heating (bovine muscle) 7 HIFU assisted PA (bovine muscle) 11 Microwave assisted PA (bovine muscle) 9 Microwave assisted PA imaging depth study (bovine muscle) 5 Microwave assisted PA imaging one line scanning (bovine muscle) 5 Differential PAR (sheep blood) 11 PVCP phantom for imaging study Light Absorption Coefficient Measurement The experimental setup for the absorption coefficient dependence on temperature is shown in Figure The prepared sample was clamped between two transparent glasses at a distance between 2 to 3 mm and wrapped in a thin plastic film. It was then immersed in a water-tank. A homemade heater raised the water temperature and the sample temperature slowly. The temperature was measured by a thermocouple (K-type, Omega). Two thermocouples were used. One was used for the sample temperature measurement, and another one was used for the water temperature. The measured temperature was read by a USB DAQ device (USB-6221, NI), and the data were sent to the computer. The 800 nm CW laser was used as a light source, and the beam size was 3.5 mm. An optical power meter (1918-C, Newport) was employed for the measurement.

40 20 Figure 2.11 Experimental setup for measuring light absorption coefficient with temperature The absorbed optical energy can be expressed with the following equation: I I exp( z) (2-7) i a where I i is the initial intensity of optical energy (W/cm 2 ), I is the attenuated optical energy intensity (W/cm 2 ), a is the light absorption coefficient, and z is the sample thickness. From Equation 2-7, the light absorption coefficient can be derived as: I ln I i a z (2-8)

41 Summary In this chapter, the instrumentation and materials preparation for FDPA experiments are described. Laser assembly, laser safety, the sample preparation process, and optical absorption measurements are presented. Emphasis is given to the introduction of the FDPA radar system.

42 22 Chapter 3 Theoretical Background of Frequency Domain Photoacoustic Frequency domain photoacoustic signal generation, Gruneisen parameter temperature dependence, signal processing, localized heating field, and the wavelength modulated differential photoacoustic radar are theoretically described in this chapter. 3 Theoretical Analysis of PA System 3.1 Frequency Domain Photoacoustic Signal Generation Frequency domain photoacoustic (FDPA) uses modulated continuous wave (CW) laser radiation, which upon absorption generates temperature oscillations in a sample and produces thermoelastic acoustic pressure oscillations. Figure 3.1. PA source geometry with light absorption The response spectrum ( r, ) of the acoustic velocity potential in an absorbing medium is generated by the spatially distributed energy source fluence E ( r, ) [J Hz/m 2 ] and can be described by the Helmholtz equation [17]: 2 2 ( r, ) k (, ) a a r E ( r, ) (3-1) c p where ka [rad/m] is the acoustic angular wavenumber, [rad/s] represents the angular c a modulation frequency, a [m -1 ] is the light absorption coefficient, [ -1 ] is the volume thermal expansion coefficient, [kg/m 3 ] is the density, c a [m/s] is the speed of sound, and c p [J/kg ] is the specific heat capacity at constant pressure. In the one-dimensional case, laser

43 23 light is incident normally to a boundary of an absorbing medium as shown in Figure 3.1. The light fluence reaching subsurface depth in the medium can be estimated by: ef E ( z, ) e z E ( ) (3-2) 0 where E ( ) 0 is the light fluence at the surface of absorber, E ( z, ) is the Fourier transform of the fluence at the specific location z of the absorber, and ef is the effective light attenuation coefficient. The effective light attenuation coefficient in the presence of scattering is defined as [18]: 3 ( ) (3-3) ef a a s where is the reduced light scattering coefficient, and is defined by: s (1 ) (3-4) s s g where s is the light scattering coefficient (m -1 ), and g is the scattering anisotropy. The anisotropy is a factor measuring the cosine of the scattering angle, and defines the scattering directional probability. Eq. (3-1) can be written in one dimension as: d dz (3-5) 2 2 a k (, ) 2 a E z cp The general solution of the second order differential equation of Eq. (3-5) is given by [17]: ( / ) ( / ) (, ) j c a z j a z C c z a e Ce E ( z, ) cp a / ca (3-6) The first two terms describe the acoustic waves traveling forward or backward inside the medium. The last term in Eq. (3-6) is the driving term. It diminishes exponentially with increasing depth and is considered as zero for 5. In the case where no wave travels toward a z the boundary medium with rigid boundary condition, so C 0, and with the rigid boundary approximation, on the surface of the medium z 0, the velocity is given as:

44 24 v z0 z z0 0 (3-7) Substituting Eq. (3-7) into Eq. (3-6), gives for C : c a a ef C j E ( ) cp a / ca (3-8) Using pressure p j, and substitutingc into Eq. (3-6), the pressure spectrum p ( z, ) can be written as: c 2 a a ( ef j / ca ) z p ( z, ) e E 0( ) (3-9) cp aca j Therefore, the amplitude spectrum of p ( z, ) can be obtained as: 2 ef z ef z acae ae p a a a a p ( z, ) E ( ) E ( ) (3-10) c c c where c c 2 a p is the Gruneisen parameter. Gruneisen parameter is a parameter influencing energy conversion from electromagnetic (EM) energy to mechanical energy. For PA phenomenon, acoustic signals can be generated by EM fluence more efficiently in medium with high Gruneisen parameter. From Eq. (3-10), we can deduce that the amplitude of the PA signal is proportional to the square of the speed of sound c a, light absorption coefficient a, thermal expansion coefficient and light fluence E 0, but inversely proportional to the specific heat capacity c p and when aca. Figure 3.2 shows the simulation of the amplitude of p ( z, ) with distance. The parameters for the simulation were as follows: frequency, 1 MHz; Gruneisen parameter, 0.15; speed of sound, 1500 m/s; light absorption coefficient, 5 cm -1 ; light scattering coefficient, 9 cm -1 ; and distance, from 0 to 0.4 meter. It can be seen that the amplitude of the pressure decreases with distance due to light absorption. The highest pressure is generated at the surface of the absorber. From equation (3-9), the phase of the pressure spectrum can be derived as:

45 25 z z cos( ) sin( ) 1 c a ac a ca tan z z sin( c ) cos( ) a ac a ca (3-11) Figure 3.3 shows the simulation of the dependence of p ( z, ) on frequency. The parameters for the simulation were: frequency from 300 khz to 1.3 MHz, Gruneisen parameter 0.15, speed of sound 1500 m/s, light absorption coefficient 5 cm -1, light scattering coefficient 9 cm -1, and distance 0.01 meter. Figure 3.4 shows the simulation results of the phase with distance. In the simulation, the frequency is 1 MHz, light absorption coefficient a is 5 cm -1, speed of sound ca is 1500 m/s, Gruneisen parameter is 0.15, and distance varies from 0 to 0.01 meter. From the simulation in Figure 3.4, we can see that phase fluctuates from π/2 to π/2 radians with the distance. The thermoelastic movements cause the phase of pressure to change with the distance. A fluctuating phase with frequency can also be obtained. Figure 3.5 shows the simulation results of the phase with frequency. In the simulation, the frequency ranges from 300 khz to 1.3 MHz, light absorption coefficient a is 5 cm -1, speed of sound ca is 1500 m/s, and the distance is 0.01 meter. Figure 3.2 Amplitude of PA signal as a function of distance (depth in the absorber)

46 26 Figure 3.3 Amplitude of PA signal as a function of frequency Figure 3.4 Phase of PA signal as a function of distance (depth in the absorber)

47 27 Figure 3.5 Phase of PA signal as a function of frequency Figure 3.6 Amplitude of PA signal as a function of light absorption coefficient

48 28 Figure 3.7 Phase of PA signal as a function of light absorption coefficient Figure 3.6 shows the simulation of PA amplitude with light absorption coefficient. In the simulation, the frequency is 300 khz, light absorption coefficient a is from to 0.1 m -1, Gruneisen parameter is 0.15, speed of sound ca is 1500 m/s, and the distance is 0.05 meter. The simulation results shows the PA signals become stronger with higher absorption coefficients. Figure 3.7 shows the simulation results of the phase with absorption coefficient. In the simulation, the frequency is 300 khz, light absorption coefficient a is to 0.1 m -1, speed of sound ca is 1500 m/s, and distance is at 0.05 meter. Figure 3.7 shows that the phase varies linearly with light absorption coefficient. 3.2 Gruneisen Parameter Dependency with Temperature Water plays an important role in bio-tissues. The total water content of the human skin is 65% and for abdominal muscles it is up to 77% [51]. The volume expansion coefficient of water increases from to between 20 and 45 as shown in Figure 3.8 [52]. The thermal expansion coefficient increases 107% from 20 to 45. Studies have shown that the specific heat capacity of water is a mildly temperature-dependent parameter, which decreases from to J/kg from 0 to 35, and then increases to J/kg at 45 as shown in Figure 3.9 [53]. The variation of the specific heat capacity of water is less than 1%

49 29 from 20 to 45. As shown in Figure 3.10, the speed of sound in water increases from 1482 m/s to 1536 m/s as temperature rises from 20 to 45, an increase of 3.6% [54]. Figure 3.8. Thermal expansion coefficient of water dependence on temperature [52] Figure 3.9. Specific heat capacity of water dependence on temperature [53]

50 30 Figure Dependence of the speed of sound in water on temperature [54] From these reference data, the dependence of the Gruneisen parameter of water on temperature can be calculated by c c p 2, and the results are shown in Figure The monotonic increase with temperature can be described by an interpolated equation from the curve in the temperature range from 20 to 45 : ( T ) T T T (3-12) where ( T ) represents the Gruneisen parameter, and T is the temperature in degrees Celsius from 20 to 45. From the results in Figure 3.11, we can see that the Gruneisen parameters of water increase with the temperature monotonically. And from Eq. 3-10, the amplitude spectrum of p ( z, ) is proportional to the Gruneisen parameter. Therefore, the amplitude of the photoacoustic spectrum increases with rising temperature and PA signal strength increases with temperature.

51 31 Figure Gruneisen parameter of water dependence on temperature In terms of molecular composition, bio-tissues are a mixture of water, proteins, nucleic acids, lipids, carbohydrates, and mineral components [55, 56]. A typical composition of bovine muscle consists of approximately 75% water, 19% protein, 3% fat, 1% fiber, 1% carbohydrate, and 1% ash [55, 56]. The composition value varies with the type of the muscle, breed method, animal sex and age, and the season of the year [57, 58]. The composition of a muscle with 81% water, 13% protein, 3% fat, 1% fiber, 1% carbohydrate, and 1% ash was used for comparison. Temperaturedependent empirical models of individual biotissue-components are shown in Table 3-1 [59, 60]. Table 3-1 Thermal property equations for biotissue components (20 T 50 ) [59, 60] Thermal Property Biotissue Composition Thermal Property Model Density (kg/m 3 ) Protein T Fat T Carbohydrate T Fiber T Ash T Specific Heat Capacity Water T T Protein c T T p Fat cp T T

52 32 (J/kg ) Carbohydrate c T T p Fiber c T T p Ash cp T T Water c T T p Composition-based thermal property prediction methods use composition data for the estimation of the thermal expansion coefficient and the specific heat capacity of biotissues. The specific heat capacity cp can be obtained from the mass weight percentage of the specific heat capacities of the tissue components [59-61]: c c w (3-13) i p p i where i cp is the specific heat capacity of a specified tissue component, and of a tissue component. wi is the mass fraction For a typical bovine muscle, the cp is calculated as: c c 0.19 c 0.75 c 0.03 c 0.01 c 0.01 pr otein water fat fiber carbohydrate p p p p p p c ash p 0.01 (3-14) where c, p r otein p c, c, c, c water p fat p fiber p carbohydrate p, and ash c p are the specific heat capacity of protein, water, fat, fiber, carbohydrate, and ash, respectively. Each value can be calculated by the equations accordingly from Table 3.1. T is the temperature in Celsius. Similarly, the density can be calculated as [59, 60, 62]: (3-15) i w i where i is the density of the specified tissue component and wi is the mass fraction of a tissue component (same as Eq. 3-13). From the density values at different temperatures, by assuming the mass is constant, then the volume change and thermal expansion coefficient can be estimated. The density of a typical bovine muscle is calculated by:

53 33 pr otein water fat fiber carbohydrate ash 0.01 (3-16) where is density and the superscript denotes the tissue component. The values can be calculated by the equations in Table 3.1. The estimated results of a typical bovine muscle in Figure 3.12 and 3.13 show that thermal expansion coefficient and specific heat capacity are all temperature-dependent parameters. The thermal expansion coefficient of 75% water composition muscle (Figure 3.12 (a)) increases from ( -1 ) at 20 to at 45, a 64% increase. The thermal expansion coefficient of 81% water composition bovine muscle (Figure 3.12 (b)) increases from at 20 to , a 74% increase. The specific heat capacity of 75% water bovine muscle (Figure 3.13 (a)) increases from 3625 (J/kg ) at 20 to 3637 (J/kg ) at 45, which is a 0.33% increase. The specific heat capacity of 81% water bovine muscle (Figure 3.13(b)) increases from 3753 (J/kg ) at 20 to 3765 (J/kg ) at 45, a 0.32% increase. The specific heat capacity of muscle tissue is 3770 to 3800 J/kg in the temperature range from 20 to 40 [61]. Figure 3.12 Estimated thermal expansion coefficient of bovine muscle as a function of temperature, (a) bovine muscle composition as 75% water, 19% protein, 3% fat, 1% fiber, 1% carbohydrate, and 1% ash; (b) bovine muscle composition as 81% water, 13% protein, 3% fat, 1% fiber, 1% carbohydrate, and 1% ash.

54 34 Figure 3.13 Estimated specific heat capacity of bovine muscle as a function of temperature, (a) and (b) have the same meaning as in the previous figure. Figure 3.14 shows the experimental setup for the speed of sound measurement. Two ultrasound transducers (V314, Panametrics-NDT) with fundamental frequency at 0.89 MHz were used for experiments. The measured sample was placed between two ultrasound transducers, wrapped by thin plastic film, and immersed in a water tank. A homemade heater was utilized to heat the water in the water-tank. Linear frequency modulated (LFM) chirp signals (0.3 MHz to 1.3 MHz, 1 ms long) were utilized as transmitted signals. The chirp signals were generated by LabView software, uploaded to the digital card NI PXI-5421, and synchronized with the data-acquisition card NI PXIe-5122 (National Instrument, Austin, Texas). The detected signals were amplified by a pre-amplifier (Parametrics-NDT, 5676) first, and then were sent to the digital data-acquisition card. Two thermocouples (K type, Omega) were used for temperature monitoring. The measured temperature data were collected by a USB DAQ device (USB-6221, NI) and then sent to a computer for analysis. The delay time of the cross-correlation peak reflects the signal traveling distance from one transducer to another (the thickness of the sample). The distance between two transducers was measured by a digital caliper ( , Mitutoyo). The speed of sound was calculated as c beef Distance delay _ time. The result for the speed of sound in water is shown in Figure Five specimens were used for the measurement. The measured results were compared to the reference values [54]. The error is

55 35 less than 0.1% from the comparison. The measured result of the speed of sound in bovine muscle is shown in Figure 3.16, where 4 specimens were used for measurement. It is seen that speed of sound in ex-vivo bovine muscle c bf varies from to 1593m/s between 20 to 45, an increase of 1.2%. The result can be fitted well to a third-order polynomial curve: c ( T ) T 0.106T T bf C T 45 C (3-17) For bovine cardiac muscle from 20 to 37 C, the speed of sound as a function of temperature is dc [63]: 1.1 dt, where dc is the speed change, and dt is the temperature change. The measured speed of sound is m/s at 20, and 1590 m/s at 37, the increase being 15.7 m/s. Figure 3.14 Experimental setup for speed of sound measurement

56 36 Figure 3.15 Dependence of speed of sound of water on temperature Figure 3.16 Dependence of speed of sound in bovine muscle on temperature The Gruneisen parameters of bovine muscle were obtained by measuring the speed of sound of bovine muscle as a function of temperature from 20 to 45C and using the estimated results of the specific heat capacity and thermal expansion coefficient. The Gruneisen parameters of bovine c muscle dependence on temperature were calculated with the equation c 2 a p, yielding the results shown in Figure 3.17.

57 37 Figure 3.17 Estimated temperature dependence of the Gruneisen parameter of bovine muscle, (a) and (b) carry the same meaning as in Figures 3.12 and The results show the Gruneisen parameter of 75% water bovine muscle (Figure 3.17 (a)) is at 22, and 0.21 at 37. Gruneisen parameter of 81% water bovine muscle is at 22, at 37 in Figure 3.17 (b). These estimated Gruneisen parameter values of bovine tissues are consistent with other findings. The speed of sound is 1578 m/s in cardiac muscle at 30 [63], the specific heat capacity of cardiac muscle is 3610 J/kg at 30 [63], and the thermal expansion coefficient of cardiac muscle is at 30 [63], so the Gruneisen parameter of cardiac muscle can be calculated as 0.21 at 30. Our estimated Gruneisen parameters are 0.16 (75% water) and 0.18 (81% water) at Thermal Field of Localized Heating Microwave (MW) and ultrasound (US) energy can be absorbed by biological tissues. The absorbed energy will transform into heat and raise the temperature in the tissues. The generated heat by MW or US is directly proportional to the absorption coefficient (MW or US) of tissue and to the frequency and intensity of the energy source [63, 65]. Heat can be reduced by increasing perfusion with temperature [65, 66]. It is assumed that the tissue samples are all electromagnetically heterogeneous and acoustically homogeneous for this study. Penne s bio-heat transfer equation was developed for roughly describing temperature elevation in biological tissues [66]:

58 38 T 2 cp k T bcbwb ( T Tb ) q t (3-18) where cp is the tissue specific heat capacity [J/(kg )], ρ is tissue density [kg/m 3 ], k is tissue thermal conductivity [W/(m )], blood density [kg/m 3 ], wb is the blood flow rate through the tissue [m 3 /(s kg)], cb is the blood specific heat capacity [J/(kg )], b is Tb is the blood temperature [ ], t is heating time [s], and q is the volume heating source [W/m 3 ]. All temperature information about blood perfusion is included in the term w c ( T T ). For ex-vivo experiments, the blood perfusion term wb cb 0. Thus, Eq. (3-18) can be simplified as [67]: b b b T t 2 cp k T q (3-19) For MW and US heating, the 3D geometry is shown in Figure 3.18, where a cylindrical coordinate system is used. The terms ( r, z, t ) and ( r, z, ) are the observation and source coordinates, respectively. The translated temperature of the thermal field in the medium is written as: ( r, z, t) T ( r, z, t) T (3-20) 0 where T0 is the ambient temperature. The translated thermal field satisfies the following equation: 2 k cp ( r ) k q( r, z) 2 t r r r z (3-21) The initial and boundary conditions for MW or US heating are given by: (r,z,0)= 0 ( r,0, t) 0 z (3-22a) (3-22b) s m q( z) q e z 0 (3-22c)

59 39 q S 0 0 (3-22d) where q0 is the heat flux at the surface z 0 and is the heat generation efficiency. Here q is the MW and US source power, taken to be only a function of the depth (axial) coordinate and uniform along the radial dimension, consistent with our experiments. The solution of the boundary value problem Eq. (3-22a) to (3-22d) can be derived using the Green Function (GF) method: 1 t b L ( r, z, t) q( z) G( r, z, t r, z, )2 rd drdz (3-23) c 0 r0 z0 p Figure Geometry of MW or US heating source For the 3D geometry shown in Figure 3.18, it is assumed that the absorbing medium is cylindrical semi-infinite lengthwise, heating spot radius is b in the radial direction, and the s penetration depth is L 1/ m in the axial direction, corresponding to the absorbing area of the MW or US energy. The GF is separated into the product of a radial and an axial function [68]: G( r, z, t r, z, ) G ( r, t r, ) G ( z, t z, ) (3-24) R A The radial function GR satisfies: G ( r, t b,0) G ( r, t 0, ) 0 (3-25) R R GR is given by [69]:

60 ( r r ) rr GR ( r, t r, ) exp I0 4 ( t ) 4 ( t ) 2 ( t ) (3-26) where I0 is the modified Bessel function of the first kind of order zero, and α is the thermal k diffusivity ( ). The axial function satisfies the following condition: c p GA ( z, t z, ) 0 z0 z (3-27) and is given by [69]: ( z z) ( z z) GA( z, t z, ) exp exp 4 ( t ) 4 ( t ) 4 ( t ) (3-28) Thus, the solution of Eq. (3-23) is: 1 t b L ( r, z, t) d G ( r, t r, )2 rdr q( z) G ( z, t z, ) dz (3-29) R c 0 r0 z0 p A This yields: 2 2 q t b 0 2 r ( r r ) rr ( r, z, t) d exp I0 dr c 0 r 0 p 4 ( t ) 4 ( t ) 2 ( t ) 2 2 L 1 mz ( z z) mz ( z z) exp exp dz z0 4 ( t ) 4 ( t ) 4 ( t ) (3-30) at r 0, z 0, the solution of the integral is given as [69]: q 0 1/2 b ( r, z, t) r0 ( t) ierfc 1/2 z0 k 4 t (3-31) where ierfc( x ) is the integral of the complementary error function. For small values of b, the solution is:

61 41 q ( r, z, t) ( ) k 0 1/2 r0 t (3-32) z0 This is the translated temperature of the thermal field in the medium with a small heat spot and constant heat flux q Signal Processing Lock-in Method Lock-in detection also called phase sensitive detection is a powerful means for recovering very small signals amongst overwhelming noise. A reference sinusoidal signal is given by: V ( t) A sin( t ) (3-33) r r r r And a detected signal is: V ( t) A sin( t ) (3-34) s s s s The detected and reference signals are multiplied by a mixer in the lock-in detector. The output of the lock-in is the product of two sinusoidal waves: V A A sin( t )sin( t ) out r s r r s s Ar As cos ( ) cos ( ) 2 t t r s r t r s r s (3-35) There are thus two frequency signals: one at the frequency r s and the other at r s. When r s, after the output signals pass through a low pass filter, the filtered output will be: Ar As VLP cos( r s ) (3-36) 2 This output signal VLP is a demodulated DC signal that is proportional to the detected signal.

62 42 Figure 3.19 shows the lock-in detection process. The detected signal Vs passes through an amplifier, and the reference signal Vr passes through an adjustable phase shifter. The two signals are mixed, and the results are low-pass filtered yielding an output signal V LP. Figure Diagram of the lock-in detection method Matched Filter Compression Method In the case of single-frequency modulation, a lock-in amplifier is used for measuring PA signals. However, the lock-in detection method cannot measure the depth of a chromophore directly [28], so a multi-frequency (sweep) approach is needed. Such an approach is offered by the chirp and pulse compression method popular in the radar field [70]. The matched-filter pulse compression method was introduced into the photoacoustic radar (PAR) system in order to give the frequency-domain PA signal necessary depth resolution with adequate SNR [28]. Matched filtering is introduced by the cross-correlation method to measure the similarity of two signals (frequency content and phase). The cross-correlation output is: R( t) sr ( t) sd ( ) d (3-37) where sd ( t) and sr ( t ) are the detected and reference signal respectively.

63 43 Figure 3.20 Linear frequency modulated input chirp waveform (up chirp). The frequency of a chirp signal increases or decreases (up or down chirp) with time as shown in Figures 3.20 and Linear frequency modulation (LFM) is the simplest and most popular pulse compression method for radar detection. It can reduce side lobes, and improve depth resolution among other attributes [28, 70]. In a chirp, the frequency band is swept during the pulse duration K. A LFM up-chirp signal can be expressed by: t B sr rect t t K K 2 ( ) cos( 0 ) K K t (3-38) 2 2 where 0 is the center angular frequency, and B is the frequency bandwidth. Figure 3.21 Matched filter response chirp waveform (down chirp).

64 44 The down-chirp LFM response signal sd is [70]: t B sd rect t t K K 2 ( ) cos( 0 ) K K t (3-39) 2 2 The output of a matched filter can be written as: B B Rt t t d K K K cos( 0 ) cos 0( ) ( ) K (3-40) 2 Eq. (3-40) can be simplified as follows [70]: t sin Bt(1 ) K t K Rt rect( ) cos( 0t) (3-41) 2 2K Bt In frequency domain signal processing, the Fourier transform of the reference signal sr ( t ) is s ( ), and the complex conjugate of s ( ) is h ( ). The matched filter output result R ( ) in r the frequency domain is obtained as: r R ( ) s ( ) h ( ) (3-42) r A simulation result of the matched filter output is shown in Figure The normalized amplitude of the chirp is 1, the frequency sweep is from 300 khz to 1.3 MHz, and the chirp duration is 0.01 s.

65 45 Figure 3.22 Output results of matched filter compression method The LFM signal s r can also be written in exponential form as: 1 t sr rect( ) e e 2 K B 2 B 2 j( 0t t ) j( 0t t ) K K (3-43) The spectrum of the LFM signal sr is: K B 2 B 2 1 j( 0t t ) j( 0t t ) 2 K K jt s r ( ) K e e e dt 2 2 K B 2 2 ( 0 ) K B j t t j ( 0 ) t t 2 K 2 K K K e dt e dt 2 2 (3-44) A variable x is defined by [71]: x ( t ) (3-45) and then dx dt, where 2 B. For the Fourier Transform of s 1 K ( ) r, we can write: K 2 K 2 B 2 j t K jt s r1( ) e e dt (3-46)

66 46 which can be written as: j j x j j x j x x2 x2 x s r1( ) e e dx e e dx e dx x1 0 (3-47) 0 where K K x1 ( ), and x2 ( ) (3-48) 2 2 The Fresnel Integrals C(X) and S(X) are defined by: 2 X y C( X ) cos dy (3-49) 0 2 y dy (3-50) 2 X ( ) sin 0 2 S X Therefore, we obtain [71]: j j x j x K x2 x 1 4 B 2 2 s r1( ) e e dx e dx 2B j 4 B K e C ( x 1) C ( x 2) js ( x 1) js ( x 2) 2B (3-51) Similarly, K 2 K 2 B 2 j t K jt s r 2( ) e e dt (3-52) after the same mathematic process as s ( ) 1, one obtains: r 2 j K 4 B r s ( ) e C( x ) C( x ) js( x ) js( x ) (3-53) 2B And the s ( ) becomes: r

67 47 1 s r ( ) s r1( 0) s r 2( 0) (3-54) 2 then: 2 j( 0 ) 1 K 4 B s r ( ) e C( x3) C( x4) js( x3) js( x4) 2 2B 2 j( 0 ) 1 K e 4 B C ( x 5) C ( x 6) js ( x 5) js ( x 6) 2 2B (3-55) where, x K ( ) (3-56) x K ( ) (3-57) x K ( ) (3-58) x K ( ) (3-59) s ( ) can also be written as: r 2 2 ( 0) ( 0) cos C( x3) C( x4) cos C( x5) C( x6 ) 1 K 4 B 4 B s r ( ) B ( 0) ( 0 ) sin S( x3) S( x4 ) sin S( x5) S( x6) 4 B 4 B 2 2 ( 0) ( 0) sin C( x3) C( x4) sin C( x5 ) C( x6 ) j K 4 B 4 B B ( 0) ( 0) cos S( x3) S( x4) cos S( x5 ) S( x6 ) 4 B 4 B (3-60) The amplitude of s ( ) can be derived as: r

68 C( x3 ) C( x4) C( x5 ) C( x6) 1 K s r ( ) (3-61) B S( x ) S( x ) S( x ) S( x ) The phase of the s ( ) can be derived as: r 2 2 ( 0) ( 0) sin C( x3) C( x4) sin C( x5 ) C( x6) 4 B 4 B 2 2 ( 0) ( 0) cos S( x3) S( x4) cos S( x5) S( x6) arctan 4 B 4 B r 2 2 ( 0) ( 0) cos C( x3) C( x4) cos C( x5 ) C( x6 ) 4 B 4 B 2 2 ( 0) ( 0) sin S( x3) S( x4) sin S( x5) S( x6 ) 4 B 4 B (3-62) Thus, the complex conjugate of s ( ) is given as: r 2 2 ( 0) ( 0) cos C( x3) C( x4) cos C( x5 ) C( x6) 1 K 4 B 4 B h ( ) B ( 0 ) ( 0) sin S( x3) S( x4) sin S( x5 ) S( x6 ) 4 B 4 B 2 2 ( 0) ( 0) sin C( x3 ) C( x4) sin C( x5) C( x6 ) j K 4 B 4 B B ( 0 ) ( 0) cos S( x3) S( x4) cos S( x5 ) S( x6 ) 4 B 4 B (3-63) The amplitude of h ( ) can be derived as: 2 2 C( x3) C( x4) C( x5 ) C( x6) 1 K h ( ) B (3-64) S( x ) S( x ) S( x ) S( x ) From Eq. (3-60) and (3-63), the phase of h ( ) is: (3-65) h r In the case when a response signal is the same as s ( ), the matched filter output is written as: r

69 49 R ( ) h ( ) cos j sin r( ) cos j sin (3-66) h h r r which becomes: R ( ) h ( ) s r ( ) cos h r j sin h r (3-67) Since h r, R ( ) h ( ) s r ( ) (3-68) And from the amplitude of h ( ) and s ( ) R ( ) h ( ) s r ( ) r, the Eq. (3-68) becomes: K C( x3) C( x4) C( x5) C( x6) 1 K C( x3 ) C( x4 ) C( x5) C( x6) 2 2B S( x ) S( x ) S( x ) S( x ) 2 2B S( x ) S( x ) S( x ) S( x ) K C x C x C x C x S x S x S x S x 8B ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (3-69) For a response signal s ( ), the amplitude is assumed to be the same as h ( ). The phase of d s d ( ) is s, which is related to the response time. s d ( ) can be written as: s ( ) s ( ) cos j sin (3-70) d d s s And h ( ) is written as: h ( ) h ( ) cos j sin (3-71) h h The output of the two signals h ( ) and s ( ) is: d 2 s ( ) h ( ) h ( ) cos( ) j sin( ) (3-72) d s h s h

70 50 The amplitude of the output is: K s d ( ) h ( ) C( x3 ) C( x4 ) C( x5) C( x6 ) S( x3 ) S( x4) S( x5) S( x6 ) 8B (3-73) Fresnel integrals can be estimated as [71]: x C( x) sin 2 x 2 for x 1 (3-74) x S( x) cos 2 x 2 for x 1 (3-75) for higher frequencies, the Fresnel Integrals are approximately equal to ½. Therefore, the final output amplitude spectrum of the two signals h ( ) and s ( ) d is: K s d ( ) h ( ) (3-76) 2B The result shows the output amplitude spectrum is proportional to the duration of the chirp, and inversely proportional to the bandwidth of the chirp. 3.5 Wavelength-modulated Differential Photoacoustic PA Signal of Differential Method In the current application two lasers with wavelengths A =680 nm and B =800 nm, are modulated by linear frequency modulated chirps at 180º phase difference. The wavelengths are chosen based on Figure The figure shows the maximum difference of the molar extinction coefficients between oxy- and deoxy-hb at 680 nm, while they coincide near 800 nm. As such, the PA signal is more sensitive in the change of blood parameters at 680 nm than 800 nm. Existing system background noise can be significantly suppressed from the system when the two modulating signals are out of phase, and the difference between those two signals would be

71 51 amplified. Therefore, the differential PA signals at two wavelengths can measure very small changes with better sensitivity. Extiction Coefficient (cm -1 /M) 3x10 3 2x10 3 1x10 3 e ox e de Wavelength (nm) Figure 3.23 Absorption spectra of hemoglobin expressed as extinction coefficient of oxy-hemoglobin eox and deoxy-hemoglobin ede [72] PA signal generation depends on both the optical and absorbing properties of the absorber. In the frequency domain, from the derived PA signal equation of Eq. (3-9), the generated PA signal can be written as: ae ( z, ) z z p ( z, ) cos sin aca aca ca ca ae ( z, ) z z j cos sin aca aca ca ca (3-77) For laser A at 680 nm wavelength, the light absorption coefficient is, and the generated signal p A can be written as: A a

72 52 E ( z, ) z z p ( z, ) c cos sin ca ca A a A A A 2 a a A 2 2 a ca E ( z, ) z A z j cos a ca sin ca ca A a A A a ca (3-78) p A can also be written as: E ( z, ) p ( z, ) cos( ) j sin( ) (3-79) A a A A 2 A A A 2 2 a ca where A is: z A z cos( ) sin( ) 1 c a a c a ca A tan z A z sin( c ) cos( ) a a c a ca (3-80) For laser B at 800 nm with light absorption coefficient at, phase shift 180º, and other B a parameters the same as laser A, the generated signal p B is written as: E ( z, ) z z p ( z, ) c cos sin ca ca B a B B B 2 a a B 2 2 a ca E ( z, ) z B z j cos a ca sin ca ca B a B B a ca (3-81) p B can also be written as: E ( z, ) p ( z, ) cos( ) j sin( ) (3-82) B a A B 2 B B B 2 2 a ca where B is: z B z cos( ) sin( ) 1 c a a c a ca B tan z B z sin( c ) cos( ) a a c a ca (3-83)

73 53 For the differential measurement, we get: 0 A B 180 (3-84) When the PA signals generated by laser A and laser B are superimposed, the differential signal p (, ) AB z is expressed by: p ( z, ) p ( z, ) p ( z, ) (3-85) AB A B So p ( z, ) becomes: AB E ( z, ) z z p ( z, ) c cos sin ca ca A a A A AB 2 a a A 2 2 a ca E ( z, ) B z z a ca cos sin ca ca B a B B a ca E ( z, ) z A z j cos a ca sin ca ca j A a A A a ca B a E B ( z, ) z B z cos a ca sin B c ca ca a a (3-86) And we can also write the differential signal p ( z, ) : AB E ( z, ) E ( z, ) p z j (3-87) A B a A a B AB (, ) cos sin A A 2 2 a a a ca A B c

74 54 Figure 3.24 PA differential amplitude dependence on two lasers power ratio Figure 3.24 shows the simulation results of the PA differential amplitude dependence on the power ratio of laser A and laser B. The power ratio P r is defined by: P r E E A (3-88) B where E A and E B are the powers of lasers A and B, respectively. In the simulation, the frequency is 1 MHz, light absorption coefficient at A B or A 1.6B 1500 m/s, power ratio is from 0.05 to 1.8, and distance is 0.01 meter., speed of sound ca is From the simulation results in the figure, we can see the differential PA signal p ( z, ) changes with the power ratio of the two lasers, and the power ratio variation can affect the sensitivity of the measurement. Thus, a proper measurement. P r value should be determined for the differential AB

75 Biomedical Application of Hypoxia Monitoring The Hb light absorption coefficient Hb is given by [73]: (,C,C ) ln(10) e ( )C ln(10) e ( ) C (3-89) Hb ox de ox ox de de where Cox and Cde are oxy- and deoxy-hb concentration, respectively and e ox and e de are extinction coefficient of oxy- and deoxy-hb, respectively. C Hb is the Hb concentration coefficient and is given by: CHb Cox Cde (3-90) Substituting, one obtains [73]: (,C,C ) ln(10) e ( )C ln(10) e ( ) e ( ) C (3-91) Hb ox de de Hb ox de ox For laser A, A =680 nm, the Hb light absorption coefficient A Hb is written: A (C,C ) ln(10) e ( )C ln(10) e ( ) e ( ) C (3-92) Hb ox de de A Hb ox A de A ox B For laser B, B =800 nm, the Hb light absorption coefficient Hb is: B (C,C ) ln(10) e ( )C ln(10) e ( ) e ( ) C (3-93) Hb ox de de B Hb ox B de B ox And since e ( ) e ( ) for the 800 nm laser, Eq. (3-93) becomes: ox B de B (C,C ) ln(10) e ( ) C (3-94) B Hb ox de de B Hb For the two wavelengths of 680 and 800 nm, the light absorption coefficient of the differential AB signal Hb is expressed as: AB A B m ln(10) e ( ) me ( ) C ln(10) e ( ) e ( ) C (3-95) Hb Hb Hb de A de B Hb ox A de A ox

76 56 where m is a constant related to the power ratio P r ( P r E E A B ) and the phase differences of two laser beams. 3.6 Summary In this chapter, the signal generation, amplitude, and phase of the FDPA signal are theoretically derived and simulated. Signal processing in the frequency domain is presented theoretically. A study of PA signal dependence on temperature, the Gruneisen parameters of water dependence on temperature, the Gruneisen parameters of bovine muscle dependence on temperature, and localized thermal field for locally PA signal enhancement are presented theoretically. Furthermore, a wavelength modulated frequency domain photoacoustic differential radar is theoretically described for PA signal improvement.

77 57 Chapter 4 Temperature Effects on Photoacoustic Radar Module In this chapter, a uniform heating enhanced method for improving photoacoustic imaging depth and signal-to-noise (SNR) ratio is presented. Experimental results showed that the maximum imaging depth was increased by 20% by raising the temperature of absorbing biotissues (ex-vivo bovine muscle) uniformly from 37 to 43, and the SNR was increased by 8%. The elements making up the Gruneisen parameter were investigated and it was found that the thermal expansion coefficient is the primary contributor to the Gruneisen parameter increase with temperature. 4 Thermally Enhanced Signal Strength and SNR Improvement of Photoacoustic Radar Module 4.1 Using Heating to Increase the PA Signal Strength and SNR The body temperature of a healthy adult is approximately 37 [33]. This value changes from person to person; moreover, different parts of the human body have different normal temperatures. The upper limit of normal body temperature is 38, and the lower limit is 35 [74]. In summary, the literature indicates that mean normal adult body temperature ranges from 35.5 to 37 [74]. Raising the temperature to 41.8 for less than an hour is generally safe for the entire human body [35]. With localized heating, most muscle tissues and skin can withstand temperature up to 44 for 30 minutes [36-38]. At higher temperature, the PA signal is stronger [42]. This phenomenon has been used for temperature monitoring [43]. In this chapter, we propose a uniform heating technique to experimentally validate the increase of PA signal strength, signal-to-noise ratio, and PA imaging depth. The response amplitude spectrum p ( ) of the acoustic pressure is generated by harmonically modulating light source fluence E ( ) [J Hz/m 2 ] in a chromophore, and can be described in one dimension by Eq. (3-10) (repeated here for convenience):

78 58 2 ef z ef z acae ae p a a a a p ( z, ) E ( ) E ( ) c c c From this equation, we deduced that the response amplitude spectrum is proportional to the Gruneisen parameter. The Gruneisen parameter of water increases monotonically from 0.11 at 20 to 0.24 at 45 (shown in Figure 3.11). Table 4.1 compares the elements of the Gruneisen parameter at different temperatures. The thermal expansion coefficient increases 107% from 20 to 45. The specific heat capacity of water is a mildly temperature-dependent parameter, with a variation of less than 1% from 20 to 45. The speed of sound in water increases from 1482 m/s to 1536 m/s as temperature rises from 20 to 45, a 3.6% increase. Figure 4.1 shows these increase rates with temperature. It is clear that the thermal expansion coefficient contributes more to the increase of the Gruneisen parameter with temperature than the other two parameters. Table 4-1 The percentage difference comparison of water Gruneisen parameter elements Water Percentage difference Gruneisen parameter % Speed of sound (m/s) [54] % Square of speed of sound (m 2 /s 2 ) % Specific heat capacity (J/kg ) [53] % thermal expansion coefficient (1/ ) [52] %

79 59 Figure 4.1 The percentage difference of Gruneisen parameter elements of water from 20 to 45. In Table 4.2, the elements of the Gruneisen parameter of bovine muscle are compared at 20 and 45. The percentage difference between 20 and 45 of the Gruneisen parameter is 67%, 1.2% for speed of sound, and 0.33% for specific heat capacity. The highest difference of the elements is 64% of the thermal expansion coefficient. The difference comparison is also shown in Figure 4.2. Table 4-2 The percentage difference comparison of Gruneisen parameter elements of bovine muscle (75% water, 19% protein, 3% fat, 1% fiber, 1% carbohydrate, and 1% ash) Percentage difference Gruneisen parameter % Measured speed of sound (m/s) % Square of speed of sound (m 2 /s 2 ) % Estimated specific heat capacity (J/kg ) % Estimated thermal expansion coefficient (1/ ) %

80 60 Figure 4.2 The percentage difference from 20 to 45 of Gruneisen parameter elements of bovine muscle comparison. In Table 4.3, the elements of the Gruneisen parameter of another bovine muscle are compared at 20 and 45. The percentage difference between 20 and 45 of the Gruneisen parameter is 82%, 2.36% of the square of speed of sound, and 0.29% of specific heat capacity. The highest percentage difference of the elements is 79% for the thermal expansion coefficient. The percentage difference comparison is also shown in Figure 4.3. Table 4-3 The percentage difference comparison of Gruneisen parameter elements for the second bovine muscle (81% water, 13% protein, 3% fat, 1% fiber, 1% carbohydrate, and 1% ash) Percentage difference Gruneisen parameter % Measured speed of sound (m/s) % Square of speed of sound (m 2 /s 2 ) % Estimated specific heat capacity (J/kg ) % Estimated thermal expansion coefficient (1/ ) %

81 61 Figure 4.3 The percentage difference from 20 to 45 of Gruneisen parameter elements for second bovine muscle (81% water, 13% protein, 3% fat, 1% fiber, 1% carbohydrate, and 1% ash) From the above comparison, the results show that the thermal expansion coefficient is the primary element of the increase of Gruneisen parameter with temperature. 4.2 Experimental Validation of Thermally Assisted PA The experimental setup was introduced in Chapter 2 (shown in Figure 2.5). A homemade heater was employed. Two thermocouples were used. One was used for sample temperature monitoring and the other was used to monitor ambient temperature. Chirp-signal modulation was implemented for the laser. A focused ultrasound transducer was served as the detector. The ink solution and bovine muscle were prepared as samples. The experimental results for PA radar signal dependence on temperature are shown in Figures 4.4 and 4.5 for the ink solution and bovine muscle, respectively. The normalized values and standard deviations (see normalization definition below) show the uniform heating effects on the PA radar signal. The standard deviations show measurement errors < 12%. It can be seen that the PA radar signals for the ink solution (about 1% liquid ink) and bovine muscle increase monotonically with rising temperature from 20 to 45. The number of ink solution samples was 7, and the number of bovine muscle specimens was 8. Each measurement was repeated 3 to 5 times. Theoretically estimated PA radar results (Eq. 3-10) were used for comparison after normalization. The normalized value n p i of the data is calculated as:

82 62 n Normalized ( p ) i pi p p p max min min (4-1) where pmin is the noise level and pmax is the maximum value of the PA signal. For the theoretical value (noise free) normalization pmin is zero. A good correlation between the estimated and normalized experimental data of the ink solution was obtained as shown in Figure 4.4. For the results with bovine muscle, the PA data exhibit reasonable agreement with the theoretically estimated values with different muscle composition (Figure 4.5). The bovine muscle composition varies with the type of the muscle, breeding method, animal sex and age, and the season of the year [57, 58]. As shown in Figure 4.5, the bovine muscle with 81% water composition exhibits good agreement between the estimated and experimental data. The Gruneisen parameter increase rate of bovine muscle with 81% water composition is 82% from 20 to 45, and the Gruneisen parameter increase rate of bovine muscle with 75% water is only 67% from 20 to 45. The results indicate that composition differences of bovine muscle significantly affect the change percentage of Gruneisen parameter with temperature. In both cases, the PA signals increase monotonically with temperature. The results are also consistent with the findings from other research on the pulsed photoacoustic signal dependence on temperature [42, 43]. Figure 4.4 PA Signal dependence on Temperature (Measured on ink solution, light attenuation coefficient μeff=3.1 cm -1 )

83 63 Figure 4.5 PA Signal dependence on temperature (Measured on ex-vivo bovine muscle, averaged light attenuation coefficient μeff=5.9 cm -1 ) compared to estimated results (bovine muscle composition). The dependence of the light attenuation coefficient of the ink solution and the ex-vivo bovine muscle on temperature were also studied. Seven ink solution samples were examined, while nine bovine muscle samples were tested. The results are shown in Figure 4.6 (a). No changes were observed in the temperature range from 20 to 45. This is in agreement with an earlier report [42]. The phase signals of the ink solutions and bovine muscles did not exhibit any temperature dependence (Figure 4.6 (b)). These signals were also measured from 20 to 45. Figure 4.6 (a) Light attenuation coefficient dependence on temperature; (b) The phase of the PA radar signal dependence on temperature.

84 64 The experimental setup (shown in Figure 4.7 (a)) for studying imaging depth improvement with increased temperature consisted of a box filled with 0.47% intralipid solution as a scatterer [75]. The tested ex-vivo bovine muscle samples were placed inside the box, and the box was moved by micrometer stages to simulate different absorber depths. Five samples were used for the experiments. In each case the foregoing heating experiment was repeated. Figure 4.7 (b) shows the experimental results of the imaging depth study at various temperatures: between 20 and 43 the maximum imaging depth was doubled, increasing monotonically from 11 to 24 mm. At 37, the imaging depth was 19mm, increasing to 22mm at 41 and 24mm at 43. The signal strength increase was about 91% from 20 to 37 and 13% from 37 to 43. The SNR increased by 52% from 20 to 43 and by 8% from 37 to 43. Figure 4.8 shows the experimental setup for PA imaging with a 64-element phased array transducer (Ultrasonix Medical Corp, frequency range from 2 to 4 MHz using elements with a pitch of mm). The laser beam was modulated by LFM chirps (0.5 to 4 MHz, 1ms long) generated by a signal generation card (NI PXI-5442). Figures 4.9 and 4.10 show the images generated by PA signals at different temperatures. At 38, a higher contrast image was generated than that at 32. The sample was a PVCP phantom. The image contrast can be measured by a contrast factor (CF) as [76, 77]: CF Signal Signal absorber background (4-2) Signal background The averaged signals were used for the CF calculation. The evaluated CF is 2.5 at 32 C for the image in Figure 4.9, and 3.2 at 38 C for the image in Figure 4.10.

85 65 Figure 4.7 (a) Experimental setup for PA radar imaging depth dependence on temperature; (b) PA radar imaging depth dependence on temperature (measured on ex-vivo bovine muscles). Figure 4.8 Experimental setup for PA radar imaging with transducer array

86 66 Figure 4.9 Image generated by PA signals at 32 C, CF = 2.5. Figure 4.10 Image generated by PA signals at 38 C, CF= Summary A study of the temperature-dependent photoacoustic signal for SNR and depth improvement was undertaken. Experimental validation was performed using the PA radar modality. The speed of sound, thermal expansion coefficient and specific heat capacity were compared as functions of temperature and were found to be factors contributing to the increase of the Gruneisen parameter

87 67 and the PA amplitude with temperature. The results showed that the thermal expansion coefficient is the primary parameter influencing the increase of the Gruneisen parameter with temperature. The experimental results confirmed the theoretical temperature dependence of the Gruneisen parameter. The study showed that the thermally enhanced PA radar can increase the signal strength 13%, SNR 8%, and the imaging depth over 20% with a temperature increase from 37 to 43 in biological tissues.

88 68 Chapter 5 Localized Heating Enhancement of Photoacoustic Radar Signals for Tissue Diagnostic Imaging In this chapter, the objective is to use localized heating for enhancing photoacoustic radar signals. Localized tissue heating generated by microwaves or ultrasound was used to improve photoacoustic radar imaging depth and signal-to-noise ratio. Elevated temperatures were experimentally measured by thermocouples in ex-vivo bovine muscle. The measured temperatures on the surface heated by microwaves or at the center of a HIFU focal area are in agreement with the theoretically estimated temperatures. The study showed that with our experimental setup microwaves can generate higher temperatures than HIFU heating in bovine muscle. Localized microwave heating can increase the imaging depth by 11%, and the signal-tonoise ratio (SNR) by 5% in ex-vivo bovine muscle. 5 Localized Heating Assistance for the PA Radar System 5.1 Introduction Thermal energy can be locally delivered by ultrasound (US) or microwaves (MW) in biological tissues [78-80]. A MW-energy applicator (antenna) is placed on the target site to increase the temperature surrounding the applicator inside the tissue. Coaxial slot antennas are the most popular tools for MW heating purposes due to their small dimensions and low cost [81, 82]. A single slot coaxial antenna was employed for this research. This type of antenna can locally produce an electromagnetic field around the slots and heat the target area (deep inside the tissue) efficiently [81-83]. US can penetrate deep inside tissues, and generate deep heating [80]. Highintensity focused ultrasound (HIFU) transducers can generate higher time-averaged intensities at their focal spot than diagnostic US transducers [80]. Focusing US energy onto a specific small volume can minimize the potential damage to surrounding tissues, because the intensities are much lower outside the focal area [78, 80]. From previous studies, the PA signal becomes stronger at higher temperatures [84, 85]. The imaging depth and signal-to-noise ratio can be improved by uniform heating [84]. Most research reports of MW and US heating were in applications of surgical ablation [65, 81-83, 86, 87]. A

89 69 theoretical comparison study of MW and US heating was conducted in the temperature range from 40 to 50 C for treating tumor cells [86]. A comparison of MW and US delivery energy differences for surgical ablation was also performed [87]. Pulsed photoacoustic systems have been integrated with HIFU for monitoring US thermal treatment or cavitation generation [88, 89]. However, low-input-power MW or US heating-enhanced photoacoustic studies for imaging purposes have not been conducted to the best of our knowledge, especially those involving the photoacoustic radar (PAR). Imaging improvement through uniform heating takes a long time and thus can cause discomfort and inconvenience to patients. Therefore, it is necessary to find a localized heating method for precise tissue targeting and reducing the exposure time. Here, we report on our investigation of heating effects on PAR signals from biological tissues through low-input-power MW or US heating. In particular, we report studies on localized MW or US heating effects on PAR signals. Tumors have higher water content than normal tissue [90] and thus have higher electrical conductivity than low water content healthy tissue [91-93]. Studies have shown that the electrical conductivity of a malignant tumor is over 10% higher than normal tissue [94-96]. This suggests that MW can raise the temperature of tumor tissues higher than the host mammary tissues, and thus generate good contrast between tumors and normal tissues. The MW attenuation coefficient is 0.09 Np/cm for bovine fat tissue and 0.81 Np/cm for bovine muscle tissue at 2.45 GHz, respectively [63, 91, 97]. The MW attenuation coefficient in other tissues has values between those of fat and muscle. The attenuation coefficient of ultrasound is also a frequency dependent parameter. At 1.1 MHz, the attenuation coefficient of bovine muscle is about 0.32 Np/cm [63]. 5.2 Localized Heating Assisted FDPA Radar by High Intensity Focused Ultrasound Figure 5.1 shows the experimental setup for the measurement of the HIFU pressure field. In this setup, HIFU was driven by continuous sine wave signals, produced by a function generator (33520, Agilent). The signals were directed through a power amplifier (50 db, ENL 240L, New York), and then through a matching network (impedance 50 Ω, Sonic Concepts) to drive the HIFU. The detector was a PVDF needle hydrophone (Precision Acoustic, UK), with a diameter of 1 mm. The detected signals were amplified by the Booster-Amp and then collected with the digital data-acquisition card NI PXIe The scanning process was started from the HIFU

90 70 surface with the step of 0.25 mm for the axial direction measurement. For the radial direction measurement, the scanning was performed perpendicularly to the HIFU axial direction (as shown in Figure) at an increment step of 0.25 mm (about half the wavelength of the 1.1 MHz signal). Figure 5.1 Block diagram of the HIFU field pressure measurement Figure 5.2 Experimental setup for HIFU heating Figure 5.2 shows the experimental setup for the ultrasound heating experiments. The HIFU transducer (H-102, Sonic Concepts) was employed as the ultrasound source. The HIFU was driven by 1.1-MHz continuous sine wave signals produced by a function generator (33520, Agilent). The signals were amplified by a 50-dB power amplifier (ENL-240L, E&I), and then passed through an impedance matching network (50 Ω, Sonic Concepts) to drive the HIFU. Temperature was measured by fine wire thermocouples (K-type, Omega). The measured data were sent to the USB DAQ device (USB-6221, NI) and then to a computer synchronized with

91 71 the function generator. The pressure amplitude of the HIFU field was calibrated with a PVDF needle hydrophone (Precision Acoustics, UK) in free-field. The pressure waveform (P) was calculated from the measured voltage and the known sensitivity of the hydrophone provided by the manufacturer. Figure 5.3 (a) and (b) show the hydrophone scanning results in the HIFU field after normalization. The measurement results in Figure 5.3 (a) show the normalized pressure distribution in the axial direction of HIFU and Figure 5.3 (b) shows the pressure distribution in the radial direction of the HIFU. The focal region is estimated by Full Width at Half Maximum (FWHM): the axial focal distance is 29.6 mm and the radial focal distance is 3.1 mm. The results showed that the axial focal region is larger than the radial region, and a cigar shape focal zone is generated. The pressure in the focal region is higher than the surrounding area. Figure 5.3 (a) Normalized pressure distribution of HIFU field in axial direction (1.1 MHz); (b) Normalized pressure distribution of HIFU field in radial direction (1.1 MHz). Ultrasound absorption is the conversion of ultrasound energy into heat, and is produced by the friction forces which oppose the motion of particles in the medium. The amount of absorption is determined by the viscosity of the medium, its relaxation time, and the frequency of the ultrasound [98]. The experimental setup diagram for the HIFU assisted FDPA system is shown in Figure 5.4. The CW diode laser emitting at 800 nm was employed. The light beam diameter was 3.5 mm; modulation was achieved with linear frequency waveform chirps (0.3 MHz to 1.3 MHz,

92 72 1 ms) generated by the signal-generation card NI PXI-5421 (National Instruments, Austin, Texas). A focused ultrasound transducer with fundamental frequency at 0.89 MHz (V314, Panametrics-NDT) was utilized as a detector. The detected signals were sent to a pre-amplifier (5676, Parametrics-NDT) and then to the digital data-acquisition card NI PXIe The size of the samples (ex-vivo bovine muscle) for US heating experiments was 50 mm long, 50 mm wide and 40 mm thick, and the size of the samples for HIFU assisted FDPA experiments was 30 mm x 30 mm x 20 mm. The samples were wrapped in thin transparent plastic during the experiments. Figure 5.4 Experimental setup of HIFU assisted FDPA system During this experiment, eleven samples were used. The FDA regulation limit for the intensity of spatial-peak temporal-average I SPTA is 0.72W/cm 2 [99]. For the localized tissue heating, the intensity of ultrasound was 3 W/cm 2 (calibrated by the hydrophone in water without an absorber), which was higher than the limit. However, temperatures of 41.8 appeared to be safe for whole-body exposure [100]. At the surrounding water temperature 37, the bovine muscle sample was heated by HIFU for 1 minute ( I SPTA = 3 W/cm 2 ) and the temperature in the sample increased to After HIFU was discontinued, in the 15 second cooling period from 41.8 to 40.5, PA measurements commenced. Figure 5.5 shows the amplitude change of the PA signal after HIFU heating. The measured PA amplitude with HIFU heating is slightly smaller than the

93 73 amplitude with uniform heating for the same temperature period. This measurement was conducted when HIFU was stopped. The HIFU was also tested during the photoacoustic measurement at the same time. However, once the HIFU was activated, the generated US signals totally dominated the acoustic field such that no photoacoustic signal could be detected by the US transducer when HIFU was operating at the same time. Figure 5.5 Comparison of uniform heating and HIFU heating effects on FDPA signals.

94 Comparison of the Heating Effects of HIFU and Microwave in Biological Tissues for PAR Measurement Figure 5.6 (a) Measured temperature on a heated surface spot of bovine muscle vs. time compared with the analytically estimated MW-heating temperature; (b) Measured temperature at the center of the HIFU focal area vs. time compared with the theory. Figure 5.6 shows the comparison of the measured temperature in ex-vivo bovine muscle of the MW-heated surface spots with the estimated temperature using Eq. (3-32): 0 1/2 ( r, z, t) r0 ( t) z0 k q where q 0 is the heating source of US or MW [W/m 3 ], and q0 is : q0 SAR (5-1) where is the density [kg/m 3 ] and SAR is the specific absorption rate [W/kg]. For bovine muscle at 37, the density is kg/m 3 [60, 84], the specific heat capacity is J/(kg ) [60, 84], and the thermal conductivity is W/(m ) [60, 63]. These parameters are all temperature dependent, but the variations are all less than 2% from 37 to 45 [60, 84], and are thus ignored in this study.

95 75 The thermal diffusivity of bovine muscle can be calculated from the defining equation: k (5-2) c p 7 This yields m 2 /s in the range The variation with temperature is ignored. For the 30-s heating in Figure 5.6 (a), all the absorbed energy was assumed converted to heat ( =100%), and the measured temperature (7 samples) at the heated surface spot was in excellent agreement with the analytically estimated temperature. The SAR was estimated to be 3.1 W/kg (the best-fit value of the theory to the experimental results), and for the heated spot size at about 2 mm, the energy density is estimated to be less than 5 mw/cm 2 (below the IEEE limit) [101]. For the 2.45 GHz microwave radiation, the MPE limit of energy density is 10 mw/cm 2 and the maximum exposure time is computed as [101]: (min) (5-3) f GHz 2.45 The differences between measured and theoretical results were less than 10%. The temperature rose to 44.5 from 37 after 30-s of microwave heating, a 20% increase (Celsius scale). No visible damage was observed after the heating experiment. Figure 5.6 (b) shows curves comparable to Figure 5.6 (a) for heating ex-vivo bovine muscle by the HIFU. A thermocouple was placed at the axial center of the focal area of the HIFU in the sample (7 samples were used). The supplied power was 2 W, the same input power as with the MW heating experiment. The ambient temperature was also 37 as well. There was good agreement between the experimentally measured and theoretically estimated temperatures during the 30-s heating (Figure 5.6 (b)). A standard deviation similar to MW-heating (error < 10%) was obtained. The estimated best-fit SAR was 0.71 W/kg and the estimated average energy density was less than 2 mw/cm 2 (spot size was about 3mm). The temperature rose to approximately 38.5 from 37 after 30-s, an increase of only 4%. Comparing Figures 5.6 (a) and (b), the results showed that HIFU heating at 1.1 MHz was not as efficient as MW heating for ex-vivo bovine muscle.

96 Localized Heating Assisted FDPA Radar by Microwave The experimental setup for microwave heating used in this work consisted of a MW generator (SSG-4000HP, Mini-circuit), a 40-dB power amplifier (ZHL-16W-43+, Mini-circuit), a coaxial slot antenna, thermocouples, and a data acquisition device (USB-6221, NI) as shown in Figure 5.7. The generated microwave signals were amplified by a power amplifier and then sent to the coaxial slot antenna. The electrical power supplied to the antenna was set at 2 W. The reflected signal was sampled at the reflected port of a directional coupler and measured by a network analyzer (E5063A, Agilent). The measured power loss was about -12 db (about 5% loss). The frequency was set at 2.45 GHz, which is the most popular frequency used for medical applications [81-83]. The coaxial slot antenna was constructed from a RG402 semi-rigid coaxial copper cable as shown in Figure 5.8. The outer conductor and core were both constructed of copper. A 2-mm ring-shaped slot was cut by removing a part of the outer conductor near the tip of the cable. The outer and inner conductors were connected at the tip by soldering. Fine wire thermocouples were employed for temperature measurements. The measured data were collected by a USB DAQ device (USB-6221, NI), and then sent to a computer synchronized with the microwave generator. The experimental setup diagram for the MW enhanced PAR system is shown in Figure 5.9. A continuous wave (CW) diode laser emitting at 800 nm was employed. The light beam diameter was 3.5 mm. Modulation by linear frequency waveform chirps (0.3 MHz to 1.3 MHz, 1 ms) was generated with the signal-generation card NI PXI A focused ultrasound transducer with fundamental frequency at 0.89 MHz (V314, Panametrics-NDT) was utilized as a detector. The detected signals were amplified with a pre-amplifier (5676, Parametrics-NDT) first, and then were sent to the digital data-acquisition card NI PXIe The coaxial slot antenna (as shown in Figure 5.8) was placed in the water tank and the slot of the antenna was placed close to the absorber surface as shown in the Figure. The frequency was set at 2.45 GHz, and generated by the microwave signal generator (SSG-4000HP, Mini-circuit). Temperature was monitored by the K type thermocouples. Bovine muscle was purchased from a local store and stored in a refrigerator at 4 C. Most experiments were performed within 3 days of the purchase. Fresh samples used for each experiment were cut from the same part. The size of the samples for MW heating experiments

97 77 was 50 mm long, 50 mm wide and 40 mm thick. The samples were all wrapped in thin transparent plastic wrap during the experiments. Seven samples were used for the temperature measurement experiments and 9 samples were used for MW assisted PAR experiments. Figure 5.10 shows the amplitude of the cross-correlation peak between the reference (by uniform heating) and detected PAR signals under the influence of MW heating. The results showed that the PAR signal strength increased about 8% with MW heating at ambient temperature 37, and the SNR increased by 5%. The average elevated temperature during the PAR measurement was approx. 42. In comparison with uniform heating, the PAR-signal-amplitude increased up to 13% from 37 to 42 with the sample immersed in heated water. For MW heating assisted PAR, the non-uniformly distributed temperature field in tissue was generated by the MW applicator, and at lower temperatures the PA signals were weaker [84]. Therefore, the non-uniform temperature distribution and the lower heating efficiency under MW heating were the reasons for the less efficient PAR signal increase. The experimental setup (shown in Figure 5.11(a)) for studying imaging depth improvement with increased temperature consisted of a box filled with 0.47% intralipid solution as a scatterer [75]. The tested ex-vivo bovine muscle sample was placed inside the box and the box was moved by micrometer stages to simulate different absorber depths. In each case the foregoing heating experiment was repeated. Five samples were used for experiments. Figure 5.7 Experimental setup for microwave heating

98 78 Figure 5.8 Geometry of microwave antenna Figure 5.9 Experimental setup of the microwave coupled PAR system.

99 79 Figure 5.10 Measured PAR cross-correlation amplitude peak from exvivo bovine muscle with microwave and uniform heating Figure 5.11 (a) Experimental setup; (b) PAR signal imaging depth study under microwave heating The results of the imaging depth enhancement study are shown in Figure 5.11(b). The sample was placed in a box filled with 0.47% intralipid solution simulating an optical scatterer [75, 84]. Five samples were prepared for the experiments. The results show that the imaging depth was increased by 11% with MW-heating assistance.

100 80 Figure 5.12 (a) Picture of a 4 mm wide ex-vivo bovine muscle sample; (b) 1-D image at 37 with optical scattering (CF=1.9); (c) Same imaging conditions as in (b) with added microwave heating (CF=2.1). Figure 5.12 shows a one line scanned image of a 4-mm wide ex-vivo bovine muscle sample. Figure 5.12 (a) is the image of the sample that was placed between clamps. Figure 5.12 (b) shows a bright PAR horizontal-line image at ambient temperature 37 with intralipid solution as the scatters. The sample was placed 18 mm below the intralipid surface. The scanning signals were collected while moving the laser beam and transducer in tandem across the sample. Figure 5.12 (c) shows the respective image generated with additional MW-heating. Comparing Figures 5.12 (b) and 5.12 (c), the MW-assisted image in Fig (c) appears in higher contrast (12% increase) than in Figure 5.12 (b) against a receding background due to the stronger signal. The CF was calculated with Eq. (4-2).