Research On Algorithm For Quantitative Photoacoustic Endoscopic Tomography

Photoacoustic tomography(PAT),an emerging medical imaging modality,combines the high resolution of ultrasonic imaging and high contrast of optical imaging.Current research on PAT includes two inverse problems,i.e.,constructing the distribution of initial acoustic pressure according to the photoacoustic signals generated by the tissues and estimating the optical absorption and scattering coefficient of the tissue.The latter,known as quantitative PAT(qPAT),is in general a nonlinear ill-posed problem.It can directly detect the physiology and pathology of the imaging structures including shape,size and optical parameters.The purpose of this dissertation is to reconstruct optical absorption coefficient distribution of biological luminal structures in photoacoustic endoscopy(PAE).Firstly,the theoretical value of the absorbed optical energy density distribution is obtained by the forward simulation based on diffusion approximation of radiative transfer equation.Then,the measured optical energy deposition distribution is reconstructed from the photoacoustic signals generated by the tissues.Finally,a squared error functional of the measured and theoretical optical energy deposition distribution is constructed and regularized.The function is then optimized with Levenberg-Marquardt(L-M)and Bregman algorithm separately estimating the absorption coefficient distribution on the luminal cross-sections.The method has been validated with computer-simulated coronary vessel phantoms including atherosclerotic plaques indicating the necessity of quantitative PAE(qPAE).The TV-and L2-regularization were evaluated according to the numerical simulation results.Comparison results between L-M and Bregman optimization indicate that Bregman optimization is more efficient than L-M in qPAE.