Study And Implementation On The 3D Reconstruction System Of Medical Images

3D reconstruction from medical images is a hot spot field. It is an important application of computer graphics and image processing in biomedicine engineering. 3D reconstruction and visualization of medical images are widely used in diagnostic, surgery planning and simulating, plastic and artificial limb surgery, radiotherapy planning, and teaching in anatomy. Study on 3D reconstruction from medical images has important significance on science and worthiness in practical application.In this dissertation, at first, the data source of medical images is discussed, after the decoding technology of CT images based on DICOM standard is analyzed in detail, pre-processing technologies including grey-scale transformation and noise elimination are presented. Aiming at the image segmentation after pro-processing, a half-interactive segmentation process is approached, which includes: creating the threshold values after analyzing the gray-level histograms, processing the images using proper mathematical morphology operation such as erosion and dilation, and drawing the outline slice by slice manually. With regard to 3D reconstruction, firstly two kinds of rendering methods including surface rendering and volume rendering are discussed and compared, and then the Marching Cubes algorithm and its problem are analyzed. Considering the speed of response, a new mesh simplification algorithm based on triangle collapse is provided and successfully used in the simplification of medical image 3D model with large number of triangles.At last, after studying the relevant excellent foreign software like 3D Slicer, we design our own Medical Images 3D Reconstruction System based on VTK. In the architecture, core algorithms are developed in C++, the visualization is implemented by use of VTK and the interface is integrated by VC++. Now several major modules have been finished, which include: data reading, image pre-processing, segmentation, 3D reconstruction, measurement, 3D operation and object separating operation.