Three-dimensional Reconstruction Of CT Images Of Brain Blood Vessels Key Technologies Research

In clinical applications, the main purpose of medical imaging is to explain and obtain high-quality medical images, At present such a problem is achieved by computer technology.In this thesis,mainly for computer technology applications in medical imaging problems encountered in the process, made a few appropriate solutions. The main work in this thesis are as follows:(1) An improved reconstruction of 3D vascular data from multiple angiographic views is proposed.3D reconstruction of the medical technology is used more dual-plane technique, but it has some errors exist, in order to reduce the error, this thesis gives two solutions, to a certain extent, to reduce error. One is the multi-view nonlinear method, the other is the multi-view linear programming method.multi-view nonlinear method through the use of multiple CT images, the RMS for the 3D distance between the center line, using the simplex method to optimize, reduce the gantry rotation and conversion errors generated in the process, can effectively avoid the double planar technology using only two views for 3D reconstruction, and can not be auto-calibration problems. Experimental results show that the results obtained using this method than using only the results obtained dual-plane technology has improved significantly.Multi-view linear programming method using linear programming methods for projection line distance between the intersection optimized to reduce the errors generated duing the conversion process. Experimental results show that the method with multi-view nonlinear method, the results have been very case, the method is faster.(2) A high contrast metal artifact reduction method is proposed.In the clinical use of 3D reconstruction images, often due to various reasons, various kinds of artifacts, seriously affected the quality of reconstruction, this thesis presents a simple method, and pseudo-first-ray coincides with part of the film through the use of the topological dual space approach to identify, and then around the part to fill a similar high-contrast target corresponding to the range. Experiment has achieved good results.