The Research Of Segmentation Method For Anomaly Area Of The Brain MRI Image

With the development of the technology in medical treatment,doctors become more and more dependent on auto medical equipment,and the achievement made in medical image technology makes the problem which is intractable in the past can be solved.In previous clinic,for medical image,doctors judge the anomaly area or lesion area only by their eyes and,manually demarcate the outline of the area.Although it doesn’t influence the carry out of medical work,but this kind of work requires by hands doctors with good background and rich experience in the clinical,The subsequent diagnose and treatment will be influence by any mistakes made by the wrong demarcation,so the traditional method cannot exclude the possibility of misdiagnose fundamentally.As a result,it is very necessary that develop the medical image segmentation technology with strong self-adaptability.Nowadays,patients have higher demand to effect of treatment,with all kinds of new illness coming,doctor’s requirement to medical equipment increases higher,especially when processing brain field image and skull field image.This article focus on a new segmentation algorithm which is aimed at medical image field.In the beginning of this article,several classic segmentation algorithms will be introduced,such as boundary segmentation algorithm and region segmentation algorithm.Which are in the early days,they can be used as medical image segmentation method.These algorithms perform well with clear and regular medical images,while they are dissatisfactory to which are not.Because the structure of human’s brain are complex and irregular,and different organizations of brain display on MRI image will show similar in their gray value,as a result when processing brain images,these classic algorithms would have significant error.Thus these methods cannot satisfy the requirement of clinic diagnose,but their appearance laid down theoretical basis and inspiration for later researchers.Fuzzy cluster algorithm(FCM)that later exploited is a great progress in image segmentation field,which can segment more complex images on the whole,and have a great improvement than early classic algorithms in self-adaptability and accuracy.But classic fuzzy cluster algorithm applied on medical image process field only solved the basic segmentation problem,the accuracy of result still need to be improved.This article emphatically introduce the improved methods that based on classicfuzzy cluster method,such as the fast fuzzy cluster method in which the Gaussian convolution is added,the improved FFCM in which the histogram equalization step is added,FGFCM in which the global space constraint model and local space constraint modelis used,FCM in which the space constraint fuzzy cluster method based on local bias estimate is adopted,sFCM algorithm in which the neighborhood impact factor is added.These algorithms which based on classic fuzzy cluster method,have improved some steps of classic fuzzy cluster method or added some steps to classic fuzzy cluster method in order to get more accurate result,stronger noise immunity and higher efficiency,and consequently achieve better segment effect.