GPU-based Improved Method Of Vascular Reconstruction

Reverse Engineering has made great progress some years age, and it is used for the medical field in this project.Project described in this paper have maken most of the 3D reconstruction of human organs and tissuesm, This paper achieve improvements in liver tissue of vascular reconstructio.Improvement focused on two areas:acceleration and smoothing. Smooth uses the heat conduction equation Smooth and VTK Smooth to make3D reconstruction; acceleration uses CUDA to accelerate MC and Heat Conduction smoothing algorithm. By optimizing of these two points, we achieve performance improvements in the three-dimensional reconstruction.The dimensional reconstruction about vessels of this paper deals with the use of the now more commonly used algorithms named MC. This method extract the vascular edge information from CT medical images of DICOM format. According to the useful information through two value processing, and converted to BMP format image, Then the extracted BMP image is used to generate the three-dimensional organ model by MC. In the three-dimensional reconstruction process, model generating quality and model building speed is the key to determining indicators.GPU-based Heat conduct algorithm save the image date that has been already processed to the GPU, then we use as parallel computing ability of CUDA to process all of the points by the heat conduction equation, so we readjust all of the bmp image points’pixel values, in order to form a more optimized image. The experiment shows that heat conduction equation has been a qualitative improvement in the speed of image smoothing by GPU.For the longstanding issues about poor imaging results and the slow isosurface extraction of Marching Cubes algorithm, In this paper, we use the very strong parallel computing capabilities of CUDA working with the CPU to extract equivalent point,and we will save the extracted data for the post processing. When we need to deal with these data-related,then we’ll call them, it avoid excessive interoperate between CPU and GPU. The experiment shows the GPU-based Isosurface extraction algorithm1 speed will make a great improve compared to conventional MC algorithm.Then we use the VTK library for further optimizing the generated results, the experiment shows that VTK make great optimization effect on the results.Finally, the project introduces the implementation of the algorithm. Software systems use the C++ programming language and C#, then we achieve the optimal of 3D vascular reconstruction,and we have developed corresponding software system.The software system is stable, and it has been certified by the CFDA.