Three-dimensional Reconstruction Techniques For Medical Image Research

Three-dimensional reconstruction of medical images transforms two-dimensional datas obtained from the CT, MRI and other medical imaging equipment to three-dimensional data in order to dislay the visual three-dimensinal form of human tissues and organs. The technology which has important practical values in surgery, tele-medicine, radiation therapy, virtual endoscopy, medical teaching has become a research focus.A anterior crucial step of medical image three-dimensional reconstruction is medical image segmentation which is the difficulty of medical image processing. This thesis reviews the existing segmentation methods, and then introduces four segmentation methods:regional growth, all interactive, live wire, Fast Marching.Three-dimensional reconstruction of medical images can generally be classified into surface rendering and volume rendering which are researched in this thesis. In surface rendering, the thesis analyzes the theory, weaknesses and improved ways of Marching Cubes algorithm which is a classic algorithm of surface rendering. To improve the surface rendering real-time interaction, this thesis analyzes the quadric error metric mesh simplification algorithm, and experimental results show that this mesh simplification algorithm can improve rendering efficiency without reducing the image quality. In volume rendering, the thesis analyzes the ray casting algorithm and presents a improved ray casting algorithm that is based ray leaping, supplemented by a variety of strategies, and achieves four volume rendering algorithms:ray casting, three-dimensional texture mapping, shear-warp, improved ray casting algorithm. Experimental results show that the improved ray tracing algorithm can improve the rendering speed under the premise of keep the reconstructed image quality.The medical image three dimensional reconstruction system named MedicalMT designed in this thesis includes five modules:image analysis, medical image pre-processing, three-dimensional image processing, interactive operation, image post-processing. The system which is successfully applied to volume and surface area calculation of glioma, segmentation and reconstruction of brain tumor, vascular extraction, skeleton extraction, detection of brain aneurysms has been recognized by clinical doctors proved that the system has some value.