Research And Realization Of GPU Based Medical Image Volume Rendering Algorithm

Visualization in Scientific Computing is a new kind of developed subject in the 1980s.It can represent large scale of data as graphics images, and has been brought into wide use. Medical image visualization is one of the research directions of Visualization in Scientific. It can reconstruct the 3D models from a series of 2D images, which can enhance the diagnosis ability of the doctors and improve the rate and quality of the therapy.Medical image mainly comes from medical image device such as CT, MRI. DICOM is an international standard in medical imaging and communications. It defines the rules of pixel data and other digital information when they are transferred or stored between devices with different operating systems or of different manufacturers. We first introduce the standard and file format of DICOM file, by making use of VS2005 and MFC, we wrote a class which can read the DICOM files.As programmability and parallel computational capability of Graphic Processing Unit(GPU) evolved rapidly these years, researches based on general purpose computation on GPU have also grew very fast, among which GPU based algorithms for 3D visualization made it possible for real time rendering on common PC. By taking advantage of hardware characteristics of GPU,3D reconstruction and image displaying using this approach can be rapidly implemented by programming through high level shading languages at lower expanse compared with the traditional method. In this article, we introduced the theory of GPU and figure out the two models of GPU:graphic API model and CUDA model.3D visualization algorithm is divided into two parts:surface rendering and volume rendering. The surface rendering extracts isosurface from 3D volume data sets and then uses computer graphical technology to display. The volume rendering takes the data as a whole to show. In this paper, In this article we put point on the volume rendering algorithm. we propose an optimization algorithm of volume ray casting using CUDA; compare the performance to previous implementation in old version of GPU. The experiments show that our method can achieve a high image quality and rendering speed.