In seventies X-rays gave the birth to radiology and thereafter radiology developed fast. Recently the invention of computerized tomography, magnetic resonance has revolutionized the radiology. By use of these technology the dream of visualizing non-invasively human internal organs and sick areas in their true form and shape has been realized. These advanced imaging methods have changed the way of traditional diagnosing and the diagnostic accuracy and effectiveness has been also improved greatly. But in terms of imaging hardware, China can not compete with the world-known imaging equipment manufacturers. However with the so-called information epoch coming, the diagnostic accuracy can be ameliorated by use of computer technology. The employment of computer technology in medicine is the trend of future iatrology. From this point of view we have studied how to exert the power of medical imaging equipment with the help of computer technology. Considering this application we have made much effort to find out the characteristics of medical images and taking these characteristics into consideration we have made lots of research of how to combine the computer technology with clinic diagnosis. Taking this point as application background, we study the problem of medical image segmentation. Segmentation is one of the most difficult problems in image analyzing and till now there does not exist a general method which can bring satisfactory result while applied to different situations. Based on the analysis of piles of literatures on this matter and considering the peculiarity of medical images, several methods are proposed in this paper. In order to provide useful information for the selection of segmentation methods it is necessary to find out the characteristics of medical images. So we have analyzed the following points: · The principle of existent imaging equipment such as CT,MRI,PET and so on. With the principle we can learn more about medical images produced using these technology. · Analyze the segmentation methods presented in literatures and point out the limitation of popular methods, and make discussion about the methodology of solving segmentation problems. Having analyzed above aspects, we come up with some segmentation methods as follows: · Proxy-based interactive segmentation method. In this paper static proxy and dynamic proxy are designed to deal with the segmentation problems. · From the point of information integration, several integration policies are presented based on texture and gray information in medical images. Sub-region-based region growing methods are proposed to carry out these policies. · Integrate the time series and dynamic programming and a so-called interactive TSDP method is given to manage the segmentation of complicated medical images. · Contribution histogram is proposed in hope for providing rich information w
修改评论