Research About Three Dimensional Reconstruction Of Medical Image Based On Volume Rendering

With the rapid development of medical imaging technology, the image data quantity isincreasing. The traditional central processor (CPU) is difficult to satisfy the clinical applicationof real-time medical image visualization demands, and the rendering speed of traditionalvolume rendering method is also relatively slow. So in this thesis, we developed an improvedRay Casting algorithm, which improves the speed of reconstruction as well as improves thequality of the rendered images.Firstly, through the comparison of different reconstruction algorithms for volumerendering, we put forward using Ray Casting algorithm as the scheme of3D medical imagereconstruction. After getting medical image data by medical imaging device and preprocessingthe obtained image data, the image quality has been improved. Then the3D image effect can beimproved. After preprocessing of the original image, we can use HL histogram transferfunctions for data mapping and Ray Casting algorithm for realizing image rendering. We alsodiscuss the basic principle of Ray Casting algorithm in detail. Through discussing the colormapping, color mixing, gradient calculation of Ray Casting，we achieved this algorithm on theGPU.According to the defect of the traditional Ray Casting algorithm that the speed of renderingis low, we put forward the scheme of improving this algorithm. Firstly, based on improving therendering speed, we accelerate the speed of the method of pre-integration. Then, considering therendering quality, we improved the gradient algorithm, estimating its gradient combined centraldifference and value function, which get quite a satisfactory result. Finally, we realize renderingimages on GPU. Through the rendering time, we can see that the improved Ray Castingalgorithm makes the speed of rendering image much faster than traditional method. Combinedrendered images and theirs corresponding histograms. We can get the conclusion that theimproved Ray Casting algorithm in rendering quality is superior to traditional method. That is the improved method not only makes the rendering speed accelerated and also improves therendering quality of medical images.At last, we shows the reconstructed images from different angles, which makes itconvenient for the doctors to diagnose diseases, and it can raise the right ratio of diagnosis.