Research And Realization Of 3D Medical Volume Data Reconstruction Algorithms

Three-dimensional reconstruction of medical volume data is an application of Computer Graphics, Scientific Computing Visualization and Image Processing in modern medicine. And it has been one of the key technologies and hot researches of Medical Visualization. To be a scientific method of medical assistant methods, it overcomes the shortcomings that doctors judge the conditions only using their experiences from reading two-dimensional slices, and can show the forms of human body and its internal organs intuitivily. The object of medical 3D reconstruction research is to provide trusty means for medical diagnoses, and to spread the new application fields as assistant diagnoses, surgery planning and training.This thesis introduces the current algorithms of medical volume data reconstruction, and studies the key techniques in it. The frame work includes that:For interpolation between slices, this thesis uses a distance interpolation method to achieve the interpolation between CT slices. This distance interpolation is based on image segmentation using wavelet transform. First, this thesis applies the wavelet transform to extract informations out of the image, then use the distance interpolation method to interpolate the goal contour. The experiment results show that this method can get the interpolation images between slices clearly.Based on studing several typical three-dimensional reconstruction algorithms and the visualization tool (VTK), this thesis analyzes the key steps of reconstruction algorithms, and simulates the algorithms using VTK, then analyzes the existent problems of algorithms and compares their advantages and disadvantages, and determines the improved direction of next research.After studying the improved ray casting algorithms, this thesis uses a template-based ray casting jump algorithm to achieve volume reconstruction of CT data. This method sets up a template at the stage of data pre-processing. While a ray arrives a transparent region, accroding to the directions of the ray, this method can determine the distance value which the ray travels to go through the transparent region, and jumps the transparent region once a time according to this diatance. This thesis uses VC ++6.0 and VTK to simulate this ray casting improved algorithm. The results show that this algorithm can reduce the rendering time and speed up the rendering process.