The Research Of Medical Image Segmentation Algorithm On Vascular Of Facial Nerve And Brainstem For Three-dimensional Reconstruction

Hemifacial spasm is a chronic disease in one side of nerve muscle involuntary, paroxysmal, painless tic characteristics, seriously affecting the patient’s life. At the same time, the facial nerve and adjacent vessels in the brain stem of complicated space structure, easy to cause the operation before the responsible vascular leakage judge or wrong, resulting in facial nerve microvascular decompression (MVD) operation curative effect is poor. In this context, this paper based on the collected MRA image, proposes the construction of facial nerve in patients with brain stem and three-dimensional structure adjacent to the vascular model. And this is the brain stem and vessel segmentation of MRA image is studied.The traditional segmentation techniques, such as single segmentation method based on single feature information or segmentation, are very difficult to obtain good segmentation results. Fuzzy clustering describes the degree of uncertainty of samples to belong to each category, closer to the attribute of things in the objective world, thus becoming the mainstream research direction of cluster analysis, changes in the level set method can automatic processing of curve topology. In this paper, the combination of fuzzy sets and the level set method, the establishment of the level set function is optimized, which makes the clustering algorithm and segmentation algorithm in speed and accuracy has been improved. At the same time, according to the characteristics of MRA images, the brain stem and vessel segmentation is performed in three steps:(1) The image clustering and brain stem extract. The imaging resolution of MRA image is not ideal, there will be interference imaging and other reasons, need for image clustering, clustering image segmentation for treatment and follow-up. Extraction based on the cluster according to the image characteristics of brain stem.(2) The segmentation of brain stem area. Early results of serial algorithm can be used for subsequent processing, has good anti-noise ability. The extraction results based on clustering and brainstem, can make a segmentation of MRA image, obtain the target region near the brain stem.(3) The extraction vessel level set. Level set segmentation algorithm is a mature image, makes the improvement to the level set, the hybrid model of boundary and region based information (DR-CV model). The DR-CV model makes full use of boundary and region information of image, can be very good to blood vessels in the area near the segmentation of brain stem segment.Proved by experiments, this method can better based on fuzzy clustering and the level set method for intracranial vascular segmentation, the ideal effect was obtained, which laid the foundation for the three-dimensional reconstruction work.The innovation of this paper:(1) By using the correlation of brainstem and vascular MRA image sequence in the space position, followed by stem and vessel segmentation. This can effectively segment the useful vascular judgment on the pathology, but also to avoid the interference of the rest of the intracranial vessels, reduce unnecessary work.(2) In order to obtain satisfactory segmentation results, with people of a variety of segmentation algorithm. This paper puts forward the fuzzy clustering and level set the combination of the two algorithms, based on the fuzzy clustering of the target image, a segmentation of the image, and then using the improved DR-CV level set model for segmentation of medical image two times, in order to achieve the intensity inhomogeneity of the segmentation purpose.(3) In order to meet the needs of medical image processing practice for segmentation accuracy requirements. In this paper, the control of the interactive segmentation method guided by users to participate in interactive method, is used to achieve the segmentation of brain stem extract.