Intravascular OCT Imaging Based On GPU Data Parallel Processing

At present,cardiovascular disease has become the leading cause of death in the world.In order to prevent cardiovascular disease early,optical coherence tomography(OCT)can accurately determine cardiovascular disease,but intravascular OCT imaging requires processing a large amount of data.According to the existing serial processing using the CPU,the data processing time is long.Therefore,the CPU serial operation algorithm is transplanted into an image processing unit(GPU)to solve the problem of data parallel processing,and image reconstruction of the blood vessel three-dimensional OCT is realized.The main contents of this thesis include the following aspects:1.A method for parallel processing of intravascular OCT imaging data is proposed.Intravascular OCT imaging technology requires processing of a large amount of data in the imaging of blood vessels,and real-time processing of 2D and 3D data by the vascular OCT system.Compare the efficiency of data processing with FPGA,DSP,GPU and CPU,and confirm that the intravascular OCT system adopts the GPU design.The data processing serial algorithm is transplanted into the GPU to implement data parallel processing of the blood vessel OCT.2.The kernel configuration and thread management optimization method of GPU parallel algorithm is proposed.The algorithm interface of GPU kernel and thread resource configuration is implemented under CUAD software system.The control interface of the GPU algorithm is outputted to the dynamic link library(DLL),and the GPU algorithm interface is investigated on the Lab VIEW software platform for simulation,and the efficiency of data parallelization is realized by comparing the algorithms of the vascular OCT imaging.3.Develop cardiovascular OCT imaging system and 3D modeling based on GPU data parallel processing method: build hardware system for intravascular OCT imaging,and call CUDA algorithm in Lab VIEW software platform to realize 2D real-time imaging of intravascular OCT imaging system.Verify the results of GPU polar coordinate calculation and dynamic thread management on the platform to achieve a larger speedup.Finally,the 3D reconstruction of intravascular OCT imaging is realized,and the imaging system of intravascular OCT is realized by GPU acceleration to pull back 300 frames of images for 3D.