Optimization Of Microwave Thermoacoustic Imaging Reconstruction Algorithm And Design Of High Performance Computing Platform

Microwave-Induced Thermoacoustic Tomography(MI-TAI)combines the advantages of microwave imaging and ultrasound imaging to obtain high-resolution and high-contrast biological tissue microwave energy absorption images.The quantitative reconstruction(qMI-TAT)algorithm can not only clearly show the structure and morphological image of the biological tissue,but also can accurately obtain the numerical distribution information of tissue parameters such as the conductivity.Therefore,MI-TAT has a broad application prospects in early detection and monitoring of many diseases.In order to obtain better results,the current qMI-TAT algorithm needs to collect enough thermo-acoustic signals generated by biological,which requires a sufficient number of large-angle ultrasonic signal sensors to be distributed around the biological tissue.However,in clinical applications,the above conditions are often difficult to meet in consideration of the hardware costs and the detection limit.Meanwhile,the existing qMI-TAT algorithm is calculated based on the Finite Element Method(FEM).In order to obtain high-resolution tissue information,it is necessary to use the element mesh as dense as possible.But that will cost more computing resources and time.In order to solve these problems,this paper studies from the following two aspects:1.Using the Total Variation Minimization(TVM)to optimize the existing qMI-TAT algorithm.Both the numerical simulation and phantom experiments results show that the optimized algorithm can obtain high quality results with a small number of sensors and limited detection angles,which will significantly reduce equipment cost and detection time.At the same time,high-quality imaging results at limited detection angles will expand the application of MI-TAT.2.A parallel algorithm for MI-TAT based on GPU(Graphics Processing Unit)was designed.Time analysis shows that the parallel algorithm greatly reduced the computational time of the MI-TAT algorithms.In addition,a high-performance computing(HPC)platform was designed,and hardware resources are integrated to achieve the goal of resource sharing and algorithm integration.Nowadays,the MI-TAT is in a critical period from laboratory to clinical application.The research in this paper can effectively improve the imaging quality and efficiency of the existing qMI-TAT algorithm.The new optimization algorithm and HPC platform are expected to lay the foundation for the further development of MI-TAT.