Research On Image Segmentation And 3D Modeling Based On Medical Image Sequence Matching

3D image modeling is a hot research field in computer graphics. 3D modeling of medical image plays an important role in medical diagnosing, radiotherapy, surgical planning, cosmetic surgery, and anatomy teaching. In the 3D medical image modeling, algorithm efficiency is a major barrier for its clinical applications.The current state of 3D image modeling research is analysed, several rendering algorithms are studied, and the ray cast principle of volume rendering is discussed. To overcome the deficiency of current rendering algorithms, some 3D modeling accelerating strategies are put forward to embed in certain key image processing routines. As the basic strategies for acceleration, the image storage mapping, the re-sampling optimization, and the sequence matching are utilized. Besides, the image segmentation is performed in order to reduce the data scale of 3D building.In the 3D modeling framework, a multi-process memory mapped file scheme is applied to accelerate image read and or write, and at the same time the medical image access is simplified. To overcome the difficulty of inside watching on any visual angle, a simulated-plane segmentation algorithm based on multi-planar incision is proposed. A similarity based sequence image matching technique is applied to increase the segmentation efficiency of large amount sequential medical images.To accelerate the segmentation process for single image, a region growth approach is performed and then a method with the combination of morphological operations and 8-neighbour contour tracing algorithm is applied to compensate the deficiency of the region growth approach. Besides, a secondary segmentation method is proposed to balance the quality and speed of the segmentation process to ensure the validation of3D modeling.Based on the above schemes, an automatic image sequence segmentation algorithm is proposed without the intervene of doctors, and based on this algorithm, a medical image 3D segmentation and reconstruction system is built. Test results show that the system is practical in the production environment.