Study On Microwave Induced Thermoacoustic Imaging Algorithm And Its Applications

Microwave-induced thermoacoustic imaging(TAI),inheriting the merits of high dielectric contrast from microwave imaging and high spatial resolution from ultrasonography,is regarded as a promising noninvasive hybrid modality for detection of human breast tumors,and other biomedical applications.The quality of images reconstructed by traditional imaging algorithm is prone to suffer from the sampling density and the widths of input microwave pulses.The denser of the spatial sampling will produce images with higher quality,but it will increase the time of the spatial sampling.The narrower of the input microwave pulse will bring ultrasonic signals with higher frequency,and images with higher quality,however it possesses less energy under the same peak power,and will degrade the signal-to-noise ratio and aggravate the demand for microwave sources with high power and narrow pulse width.For the issues aforementioned,In this dissertation the compressive sensing applied in the reconstruction of TA images is used to study about the effect of the input microwave pulse on the quality of the reconstructed images when spatial undersampling is carried out,then the deconvolution combined with the BP algorithm is applied to study about the impact of the input microwave pulse on the quality of the reconstructed images.In addition,TAI is applied to detect the explosives embedded in high-water content,and the impact of microwave frequency,embedded depth and the thickness of the explosive on the detection are studied further.The main contribution of the dissertation is as follows:1.The thermoacoustic(TA)governing equations are derived based on physical procedure of the microwave TA effect and principle.The analytical expression of TA equations is derived by introduction of the Green's function under the condition of the homogeneous acoustic and thermal properties.The analytical solution is applied to investigate the acoustic waves from a spherical target illuminated by a rectangular pulse,then the impact of the radius of the target and the width of the input microwave pulse on the generated acoustic signals.In addition,the TA dictionaries used in the imaging algorithm of compressive sensing can be built up extremely quickly and accurately by the analytical solution.2.Due to the analytical solution of TA equations is not available for inhomogeneous acoustic and thermal medium,the numerical solution of the whole microwave TAI is built up,and two numerical approaches in time domain,FDTD and PSTD,are applied to solve TA equations.Compared with the spatial derivation implemented by the local central difference in the FDTD approach,the PSTD approach applied global derivation based on FFT to realize the spatial derivation,which reduces the number of elements,and improves numerical efficiency largely.The effectivity of FDTD and the efficiency of PSTD approaches are validated by the 2-D homogeneous,3-D homogeneous and inhomogeneous cases.3.Due to the densely spatial sampling of the traditional imaging algorithms,the TAI algorithm based on CS is studied.Firstly,the area to be imaged is divided into many small square elements,and the TA analytical solution is applied to calculate and record the TA signals from every cell at every transducer.Then the procedure of TA image reconstruction is converted to solve a convex optimized problem.Secondly,the relationship between the microwave pulse envelope and the quality of images reconstructed by TA algorithm based on CS is studied when spatial undersampling is carried out.The images reconstructed by CS possess higher peak signal to noise ratio(PSNR)than those by BP,and are robust for different microwave pulse envelopes.4.Regarding the linearity of the TA signals in terms of the microwave pulse envelopes,the deconvolution of microwave pulse envelopes by conventional Fourier transformation is prone to suffer from the systematic noise.The BP algorithm combined with deconvolution of microwave pulse envelopes based on generalized cross validation and iteration is proposed to reconstruct images.The numerical and experimental results show that the images reconstructed by the hybrid algorithm possess high PSNR,and PSNR and the correlation coefficient are constant for various shapes of microwave pulses.5.Detection of explosives embedded in a high-water content medium(known as contact detection),such as mud,meat and animal carcasses,is a quite natural application for TAI because bulk explosives and the background medium have large microwave property contrast in terms of both their dielectric constant and conductivity.Firstly,the TAI approach is applied to study on detection of explosives embedded in a high-water content medium.The TAI modeling suitable for the explosive detection is proposed,and the effectivity of TAI for explosive detection and the accuracy of the numerical modeling introducering the property of the spherical focus transducer are validated both by simulations and experiments.Secondly,the microwave frequency,embedded depth of the explosive in the medium and the thickness of the explosive is studied systematically through simulations.The relationship between detection depths and input microwave power at different frequencies are given which provides insight on appropriate selection of microwave frequency and input peak power for applications of TAI on explosive detections.Lastly,an abstraction of the non-contact detection of the explosive through TAI is verified by an experiment in which vibration induced by a transducer is detected by a W-band vibrometer.