Research On Segmentation Algorithm Of Hippocampus In Brain MRI Image

With the development of discipline-medical image processing and analysis, many new methods have appeared in the field of image segmentation for medical image. But there is no universal theory and method to segment different object. Many of these image segmentation methods are used for specialized objects. The segmentation effect may not be the ideal one when applied to other objects. Furthermore, since the hippocampus segmentation in brain MRI image is the key and foundation of volume measurement and3-D reconstruction of hippocampus formation, the volume measurement of hippocampus provides an important high-efficiency method for some medical image diagnoses, such as temporal lobe epilepsy, Alzheime’s disease and schizophrenia. Therefore, the research of hippocampus segementation in brain MRI image is of great significance.The brain hippocampus segementation of the MRI image has been studied, including algorithm design, evaluation and the construction of the platform. Based on the comprehensive introduction and analysis of multitudes of existing medical image segmentation, the physiological anatomical structure of the brain hippocampus and the characteristics of MRI have briefly been analyzed; Secondly,according to Snake model and gradient vector, the hippocampus MRI image segmentation method is put forward based on GVF Snake model, which has solved the problem of concave edge detection; furthermore, for the low speed of parameters active contour model and the sharp sensitive to initial contour, the segmentation method of the hippocampus MRI image, which is based on geometric active contour model, is proposed; The traditional level set method and classic C-V model is compared thoroughly. Finally, in order to solve the problem of unprecise segmentation of the traditional level set method based on edge information and C-V model based on regional information, the level set method based on the region and edge information is presented, which have achieved good segmentation effect with improved segmentation rate; As for the performance evaluation of algorithm, commonly-used evaluation methods have summarized, and the comprehensive evaluation, which is suitable for the segmentation of hippocampus, has been put forward. It is proved to be of rationality and reliability by experiments. In the platform construction, the design of the segementation platform based on this paper is introduced thoroughly, and proved to be of feasiablity, meanwhile, the accuracy and reliability of the algorithm in this paper is indicated by the experimental results.