The Research Of Metal Artifacts Segmentation Techniques In Medical Image

With the development of Computer Generated Imagery technology, medical image processing and analyzing technology has been more and more important. But medical image and image in real life have differences, performance in the poor uniformity, gray fuzzy and other uncertainty. When the patient carries a metal object and can't take off it within his body during the scanning, obvious metal artifacts will emerge. These abnormal images can reduce the images' quality; even affect the analysis and diagnosis of certain pathological changes. That's why we have to research on how to remove metal artifacts and put it into practice whether in terms of diagnosis or radiotherapy positioning. So the image segmentation technology plays the key role in removing CT image artifacts. The main contributions of this paper are described as follows:(1) Classifying medical image artifacts and study the basic principles of revise metal artifact.(2) Introducing the threshold segmentation of image area information oriented:the iteration method of optimal threshold and OTSU threshold segmentation method. We select four typical image simulated. The experimental results show that optimal threshold segmentation is better than histogram segmentation(3) Analyzing the method based on edge segmentation:Prewitt template, Sobel template, Laplace operator and Canny operator. Four typical images will be simulated. It indicates that unreasonable outline will generate due to the enhanced noise within close tolerance for edge detection operator. Operation time decreased while the partition result generates deviation. Meanwhile morphology operation has been raised. Chief applications of morphology are edge detection and region filling. The results demonstrate that medical image segmentation effect is preferably with the metal artifact, image distortions occur without image segmentation while operating efficiency has been enhanced.Based on region growing algorithm analyzed mountain climbing algorithm and watershed algorithm. The advantage of mountain climbing algorithm is that the edge to center growth pattern can avoid excess without the threshold selection. On the contrary, watershed algorithm can obtain an excellent segmentation effect at an interface of closed gray.(4)Based on one dimensional OTSU threshold segmentation method, two-dimensional OTSU threshold segmentation method is proposed and achieved superior image segmentation results.(5)Putting forward a threshold segmentation algorithm based on the concept of information entropy to break up medical images. This kind of method determines the segmentation threshold through the entropy information of gray distribution. Nowadays, the threshold segmentation algorithm based on the concept of information entropy receives its development. The new gray gradient of two-dimensional histogram is better than traditional histogram. It considers the background of interior point and the target of interior point comprehensively, improve the effect of segmentation, reduce the space of traverse the solution. Due to the complexity of medical images, this algorithm can improve efficiency. The way of entropy Tsallis which prove as an effective method. Compared with other segmentation methods, it has a strong applicability and the results are good--it can split metal parts.