Algorithm Of Endoscopic Photoacoustic Tomography For Acoustically Heterogeneous Tissues

Biological photoacoustic tomography(PAT)is a newly emerged multi-physical functional imaging modality.The acoustic heterogeneity of biological tissues with different chemical compositions is usually difficult to be measured before imaging.It causes the acoustic reflection,scattering,diffusion and absorption when the ultrasonic waves propagate in the tissues,leading to the attenuation of the photoacoustic(PA)signals.The PA signals with higher frequency may attenuate more seriously.Consequently,the PAT images reconstructed with the standard image reconstruction methods usually contain blurring,artifact and distortion caused by the ideal assumption of acoustically homogeneous medium with non-attenuation and constant speed of sound.Both the image quality and imaging depth are reduced.The purpose of this thesis is to design image reconstruction algorithms for endoscopic PAT(EPAT)in acoustically heterogeneous tissues.The reconstruction quality is improved by compensating for acoustic scattering and attenuation in the tissues with heterogeneously distributed speed of sound.The main work includes two aspects.First,the Green function in the scattering medium containing the scattering domains is reconstructed.The images of the initial acoustic pressure distribution on the cross-sections of the cavity are then reconstructed with the time-reversal(TR)method to reduce the scattering artifacts.Second,the correlation equation of the attenuated PA signals and the ideal signals is obtained and discretized.The equation is then solved by utilizing the singular value decomposition method(SVD)to obtain the approximation of the ideal signals from the measured signals.Finally,the images with reduced artifacts are reconstructed from the ideal PA signals by using the standard TR method.The validity of the algorithms was demonstrated with computer-simulated arterial vessel-mimicking phantoms.Comparison experiments with other methods were performed to demonstrate the superiority of the proposed algorithms.The results show that the Green-based method and TR adaptive focusing(TRAF)method can effectively reduce the scattering artifacts and target dislocation in the reconstructed images.However,the TRAF method needs the prior knowledge of the distribution of acoustic scatterers in tissues,which limits its applications.The Green-based method can solve the problem of image reconstruction in the case of unknown distribution of acoustic scatterers.The SVD-based method outperforms the method of correcting the attenuation of different frequency components(abbr.correction method)in effectively compensating for the acoustic attenuation and improving the focusing of the image.However,the correction method can only compensate for the exponential acoustic attenuation,and it needs the prior knowledge of the attenuation coefficient of tissues.