Research Of Approaches In Medical Image Segmentation Based On Level Set

Image segmentation is to separate an image into a lot of un-overlapped and homogeneous regions. As a fundamental technique of medical image processing, it is an indispensable important step in the process of qualitative or quantificational analysis to human organization.The level set method based on geometric deformable model, which translates the problem of evolution of 2-D(3-D) close curve(surface) into the evolution of level set function in the space with higher dimension, obtains the advantage in handling the topology changing of the shape. As a accurate and steady algorithm, the level set method has a wide application. However, as the traditional level set method just using the local marginal informations of the image, it is difficult to obtain a perfect result when the region has a fuzzy or discrete boundary, and the leaking problem was inescapable appeared. In this paper, a level set segmentation algorithm based on Bayesian classification for medical image segmentation was proposed. Firstly, the probability that level set curve lies on boundary was estimated by using Bayesian classification model; Secondly, the term of region strategy, which is correlative with the forenamed probability, was appended to the level set function; Finally, by using the images regional information, we improve the level set method on identifying the boundary. Results of the segmentation tests prove the proposed segmenting method preferably solve the leaking problem in medical image segmentation.Additionally, in the course of calculation for the traditional level set method, we usually use first difference to approximately substitute the differential coefficient of the functions. However, with the influence of image degeneration caused by many factors such as noise, it often obtain a bad segmentation result. A level set method utilizing Hermite derivative filter for medical image segmentation was proposed in this paper, which made the calculating process to be more accurate. It was prove thatthe algorithm utilizing Hermite derivative filter has a better segmentation result than that of the traditional level set method, especially when the image was interfered by noise, the proposal has a special superiority. At the same time, the new algorithm will not increase the cost-time of the segmentation process.Lastly, we introduce a level set PDE based on the simplified Mumford-Shad model for image segmentation proposed by Chan and Vese, which shows less insensibility of initialization and noise affect, and has the ability of detecting both inner and outer edges of targets with inner hole just by one enclosed active contour, even the region has a fuzzy or discrete boundary. However, the edges far way from the active contours would be seriously suppressed by the Dirac function in the proposed PDE, and it will cost longer time to accomplish the segmentation process. So, finding the optimization algorithms will be the main aim in my future research.