Research On Photoacoustic Imaging Algorithm Based On Fast Regularization

Photoacoustic imaging is an emerging biomedical imaging tool that has the advantages of both optical imaging and acoustic imaging.Compared with optical imaging its imaging depth and high-resolution characteristics,compared with acoustic imaging have the advantage of the functionality and it is a non-invasive,non-invasive,non-ionizing means of imaging.In photoacoustic imaging,the whole physical process is a pathological problem due to the limited monitoring data and the influence of noise.The best way to solve the problem is the regularization method.Among the traditional regularization algorithms widely used is the Tikhonov regularization scheme.However,due to the requirements of imaging accuracy and imaging efficiency,the size of the solution is usually very large,which causes the burden of imaging time.To this end,this paper develops a fast regularized imaging algorithm that solves the inverse problem of photoacoustic imaging quickly using the compressed perception method for fast regularization of photoacoustic imaging in Euclidean space and the Lanczos iteration method,respectively.In this paper,we first develop a method for optimizing the regularization parameters of photoacoustic imaging based on the generalized Lagrange full variational method to solve the regularization parameters in Euclidean space quickly,which can solve the photoacoustic regularization parameters quickly and give the results of the inverse problem.And it has been verified.Another work in this paper is to identify the shortcomings of the existing LSQR fast regularization scheme(regularization method in Krylov space)and improve it.The existing LSQR scheme suffers from the problem that the objective function does not fit into the Krylov space,there is no screening mechanism for the regularization parameters in the Krylov space and the problem of the dimensionality of the Lanczos iteration termination condition,i.e.,the Krylov space.Based on the three problems,MINRES type objective function is proposed respectively,and the scheme of KRLODPLSMR-GCV3 DC based on the principle of de-left orthogonal matrix difference in Krylov space and the method of GCV plane curve family is combined into KRLODPLSMR-GCV3 DC for experiments.In this paper,matlab is used as the experimental platform to conduct experiments for the data of biological tissue detection signals.The experimental results show that the method is effective in reducing imaging time and improving imaging quality.