Key Techniques Of Virtual Endoscope Based On Pc

Virtual endoscopy (VE) is a typical medical image processing technology,which relates to computer science, graphics, data visualization, anatomy,radiology and many other fields. Virtual endoscopy is relative to the traditionaloptical endoscope, which uses image processing technology to reconstruct thehuman organ endoscope perspective image similar to the traditional opticalendoscopy from a set of serial images, is an emerging medical aided diagnosistechnique. Compared with the traditional optical endoscopy, it is a noninvasivediagnostic technique, which has many advantages, such as no pain, no infection,permanently stored data, and it also has great prospect.In this thesis, we conducted thorough researches on VE, mainly includingtomographic image preprocessing, medical tissue extraction, virtual sliceextraction and3D rendering algorithm. In the tomographic image preprocessingpart, we firstly use the Gauss filter to process the image, and then we useHermite interpolation, in order to solve the inhomogeneity problem of the slicesbody density in x, y,z directions. Due to the different bone gray value causedby the different bone density in the same slice, we use the3D generalizedgradient vector flow image segmentation method based on the Snake model tosolve that problem during the medical tissue extraction part. Most of the currentmedical imaging equipment is only along a fixed direction to produce series ofimages, the number of the slice is fixed and finite too, if you want to havedifferent directions of the slice, you would have to scan again. Therefore, thisthesis proposes a3D arbitrary angle virtual slices extraction algorithm based oninterpolation. Finally, we use three-dimensional surface rendering techniques todesign the interactive skull virtual endoscope system with the Matlab GUIinterface, the system includes3D interaction, virtual slices displayed andscreenshots and other functions.