Research And Implementation Of Visualization Technology For Cerebrovascular Medical Image

3D reconstruction from medical images is a multi-discipline subject, which is an important application of computer graphics and image processing in biomedicine field. It can process3D Volume Data which comes from2D digital tomographic images, and transform the medical image data to3D appearance; this technology can provide more useful visualization information for clinicians. It is not only important to theory research, but also valuable and helpful for clinic diagnosis and so on.There are two different approaches of3D reconstruction from medical images, including surface rendering and volume rendering. This thesis briefly researches several typical methods of surface rendering algorithm and volume rendering algorithm, and analyses their advantages and disadvantages. Compared with surface rendering in3D reconstruction, volume rendering can concentrate on visualizing internal features of choice and surfaces at the same time. Volume rendering is the emphasis of this thesis.In this thesis, the characteristics of the CT or MRI image data and preprocessing methods are elaborated. The thesis analyses the principles and differences of different visualization methods, and puts emphasis on classical ray casting and splatting volume algorithms for three-dimensional reconstruction process, and particularly analyses the basic principle of ray casting algorithm, then deeply researches the key techniques such as data classification, distributing color and opacity value to each voxel, resample and trilinear interpolation, phong illumination model, shading calculation and image composition, etc. Aim to the problems of high frequency information loss, univariate opacity assignment, large computation and slow rendering velocity, this thesis proposes a feasible improved algorithm which changes the sampling sequence, distributes opacity using multiple factors, resamples background voxels in leaps and bounds; also particularly analyses the basic principle of splatting algorithm, then deeply researches reconstruction function and footprint lookup table. This thesis designs an accelerated algorithm using similarity of successive medical images, which calculates the contribution of pexels on the screen by the density subtraction between adjacent layers to reduce the time. Finally, the visualization software MITK is introduced, and the thesis completes several different visualization methods using MITK, achieves a cerebral vascular MRI volume data visualization image, then validates the correctness and high efficiency of the algorithms, and obtains the anticipate results.