Research Of Medical Image Segmentation Based On Adaptive Threshold And Markov Random Fields

In recent years, biomedical imaging technology has been developed rapidly and a large amount of high-resolution medical image data have been used for people. A variety of medical imaging technology has been widely used in medical diagnosis, pre-operative plan, the treatment of postoperative monitoring etc. It is essential that these various imaging techniques to obtain quantitative and qualitative analysis of the data so that these valuable information comes into fully effective usage. Therefore, the segmentation of medical technology as medical image processing and analysis is a key issue. And segementation and identification of Regions of Interest (ROI) in the image is the key problem, which is the bottleneck that restricts in development and application of other medical imaging technology.Image Segmentation algorithms based on adaptive threshold and Markov random fields are researched in this thesis. The main work is as follows:1) Two kinds of segmentation algorithms based on adaptive threshold for Single frame of the medical image are researched. In response to low-contrast, low-noise ratio, non-uniform brightness of the medical image segmentation, and analysis of the causes of non-uniform brightness, the two algorithms are presented. One is based on neighborhood information to calculate the threshold surface, and the other is to use the edge information to calculate threshold surface of the adaptive local threshold segmentation, two approaches can solve the problem of segementation error caused by uneven lightness in the image.2) A segmentation algorithm based on adaptive threshold for sequence of medical images is presented. The differences of inter-frame and with background are analysed, and segmentation method based on differences between target and background is researched. Three methods to generate background are compared and analysed, and a self-adjusting threshold selecting method based on the quick Euler number to distinguish target in background is proposed.3) Segmentation method for medical image based on the MRF is researched, and an improved image segmentation algorithm based on the MAP MRF is presented. Experiment results show that the algorithm can extract Region of Interest (ROI) in medical image effectively.