Research On Algorithm For Medical Image Segmentation Based On Level Set

Medical images is important for doctors to diagnose the disease, good medical imagesegmentation can help doctors quickly determine the patient’s condition and to takemeasures to make good the patient back to health. With the rapid development of computergraphics technology, computer technology medical image processing research in recentyears become the focus of a wide range of medical image segmentation algorithm in thecontinuous research, but no one algorithm can handle all medical images. Level setalgorithm as earlier proposed algorithm achieved a more satisfactory development inmedical image segmentation. The level set by the energy function correction algorithm forthe original function, approaching object contours to be segmented. Currently, thisalgorithm generally edge detection and region-based information to establish an energyfunction.With the traditional distinction between the most significant medical imagesegmentation image segmentation is divided randomly distributed in medical image noise,due to the complexity of medical images, the traditional segmentation method is likely tocause a segmentation fault, using the level set and reasonable adjustments parameters, youcan achieve better segmentation results. However, the local edge traditional level setmethods used for medical image noise and edge discrete parts division will generate anerror, it is difficult to achieve good segmentation results. Medical image segmentationprocess is similar to the human eye of the objective world to classify different objects,which from the image of the associated structure (or ROI) isolated, image analysis andrecognition is the primary problem, but also the medical image processing constraints otherbottlenecks in the development of related technologies and applications. Therefore, medicalimage segmentation is an important image processing technology, automatic image patternrecognition, scene analysis and processing steps to understand a crucial deal from low-levelto high-level image understanding bridge.This paper describes the theory and related medical image segmentation methods, andlevel set image segmentation algorithm are compared and analyzed, focusing on thesegmentation algorithm based on a variety of segmentation level set method and the currentlevel set low. This paper presents a method that first fuzzy clustering method to improve theinitial image processing, we can get better results than the traditional method of fuzzy clustering fuzzy clustering method improvement. Divided by the initial results obtainedusing the level set method for a second more detailed segmentation, using the penaltyfunction to avoid the re-initialization of the level set by the rules of the distance of the levelset evolution to improve, get less computation and better segmentation results.In our experiments, we used the MATLAB R2010B simulation experiments, the levelset segmentation results for a variety of contrast through the use of CT images, and theresults of the algorithm is a detailed description and analysis. The results show that thealgorithm to obtain a good segmentation.In this paper, based on the level set segmentation method were analyzed and comparedto make meaningful improvements. However, since the image segmentation involves avariety of factors, such as image quality, the accuracy of the initialization, the main imagecontains a lot of knowledge of the relevant art, is limited to the current segmentationalgorithm, and an algorithm cannot all images get a good segmentation. Future mightconsider using artificial intelligence methods for image segmentation operation, because inaddition to containing the image pixel information, and more cross-cutting areas ofknowledge included, if we can cross the field of knowledge applied to the imagesegmentation, may get more good results. So how to get a better quality of medical imagesegmentation remains a challenging problem, people need more in-depth discussion andresearch.