Application Of RSF Algorithm Based On Clustering In Medical Imaging Segmentation

Image processing is an important computer technology, which extracts the original image information, and refine for visual elements, with practical value was widely attention of the people's livelihood and technology applications. The continuous improvement of people's living standard, makes the role of medical imaging appeared gradually. The development and improvement of medical imaging technology make the very big enhancement and development space for the medical image processing, and the processing is a modern science and technology with great theoretical and practical.But features inherent in medical image such as artifacts, noise, boundary fuzzy and intensity inhomogeneity, also increased the difficulty of visualization and adjuvant therapy. The difficulties of medical image segmentation is:gray, edge and color features exist in medical images, and the traditional image segmentation algorithm is not ideal, so medical image segmentation has become a research focus, hotspot and difficulty.Single traditional algorithm have been unable to meet the needs of medical image segmentation. To address the problem, FCRSM (Fuzzy Clustering and Local Robust Statistics-based LS Model) is proposed in the paper, combining with Clustering and RSF Model in the PDE equation framework, to obtain improved robustness and accuracy of image segmentation technology. In this paper, the core is to construct the fusion model of the fuzzy set theory and deformation model: traditional k-means clustering transition to the fuzzy c-means clustering, redefining the grayscale medical image pixels with fuzzy membership degree, and introducing relevant entropy and local robust statistics, enhance image robustness to noise and artifacts. The core of the algorithm is level set active contour model in deformable model, RSF algorithm, and the improved level set FCRSM model is designed, with the integration of fuzzy clustering and statistical information.Traditional RSF model, combined with the improvement RSF model of k-means-CRSM algorithm and FCRSM methods, is compared in experimental section respectively, which implemented qualitatively and quantitatively. The results shows that FCRSM algorithm is better than the traditional level set method in segmentation accuracy and effect under the reasonable size of resolution.The main work of this paper:(1) the type of image segmentation, and some typical image segmentation technology and its improvement is described.(2) the image segmentation application of k-means clustering and level set algorithm is briefly described:using simple and efficient characteristics of k-means clustering, make k-means with the combination of level set; Through the comparison of the improved CRSM algorithm based on robust statistics, the advantages and disadvantages is analyzed.(3) aiming at the shortcomings of the traditional RSF and CRSM algorithm, combined with the fuzzy c-means, the improved FCRSM algorithm is worked, its application in medical image segmentation and the advantages and disadvantages is studied, integrating into the RSF level set model, to obtain higher processing accuracy and noise robustness.(4) the traditional segmentation technology and improvement is showed in quantitative and visual contrast, and algorithm improvement is analyzed. Innovation of this article is that a new model is formed, with integration of the fuzzy clustering and local statistics, the entropy and level set model.This article research significance:make medical image segmentation promote high-level of image analysis and understanding better; Also for image registration and fusion, identify services, to promote medical value. Allow doctors to combine professional medical knowledge, with test results, to quantitative diagnosis, drug curative effect evaluation and prognosis prediction, to develop more scientific and reasonable therapeutic regimen.