The Research Of Medical Image Fusion Algorithm Based On Level Set Regions Segmentation

Medical image fusion technology is a hot topic research at home and abroad, it can achieve more comprehensive and better images by integrating the complementary information of medical images, which can provide more accurate data for clinical diagnosis and treatment. How to use the information provided by various types of images effectively to get the best fusion image has important theoretical and practical significance.The traditional medical image fusion methods have the problem of not fully considering the target characteristics of image, target to the problem, we introduce the medical image fusion method based on level set region segmentation, make a thorough study and do a large number of experiments in multi-resolution analysis, segmentation and analysis methods, image fusion strategies based on region regmentation according to its characteristics. Especially focusing on active contour model based on level set segmentation method for medical image-based regional. In this paper,we choose the CT and MRI images as objects of study, thesis researches are as follows:1. The level set model based on the boundary is sensitive to noise, it is often unable to deal with uneven regions and sensitive to the initial coutour, therefore, we proposed a medical image segmentation method based on improved active contour model and visual features, using single level set model to segment multi-phase image by introducing marked matrix and getting the initial contour based on saliency map extraction method, which can improves the accuracy and speed.2. The level set model based on the regions, which is hard to segment the complex multiphase image, and the problems of inaccurate positioning and over-segmentation in the segmentation process are very common.At the same time, the level set model convergents slowly and is sensitive to the initial contour, for these problems, an segmentation method based on improved multi-phase Chan-Vese level set model and regions of interest is proposed. The model gradient information, edge detection operator and the penalty term were introduced into the energy function, and the initial contour was obtained by FCM method , the experimental results show that the proposed method can effectively deal with medical images with lesions.3. The traditional pixel-level image fusion method in the integration process did not fully consider the target characteristics of the images, the general fusion algorithms based on region segmentation are lack of relevance in the regional division, based on this, we proposed a new image fusion algorithm based on NSCT and level set segmentation, which divided the image into regions of interest and background based on saliency map and fused the coefficient according to the regional division situations. Experimental results show that the fusion image inherits the target information of source image well with more clarity.