Extraction Method Of Brain Midsagittal Plane In CT Images

The midsagittal plane(MSP)of the brain can be used to assess the symmetry of the left and right hemispheres of the brain.The results obtained can help diagnose some brain diseases.Therefore,the study of MSP extraction is of great clinical significance.However,most of the existing extraction methods have problems such as high computational cost and certain requirements on image orientation.For these problems,the paper studies an algorithm suitable for brain images in all directions,and can quickly and accurately extract the midsagittal plane in the brain.According to the dimensions of the processed objects,MSP extraction methods based on brain symmetry can be divided into two categories.The first type of method is to perform multi-scale search on three-dimensional images.This method has no limitation on the initial direction of CT images of the brain,and emphasizes the overall symmetry of CT images,so the results will be deviated when they are refined to each slice.In addition,due to the need for multi-scale search,the cost of this type of method is expensive.The second type of method is to search for two-dimensional images.This kind of method can effectively overcome the above shortcomings,but it is necessary to ensure the existence of symmetry features of each slice,so there is a certain requirement for the initial direction of CT images of the brain.The midsagittal plane extraction algorithm proposed by the thesis is an improvement of these two methods.The main innovation is that,combining the advantages of the above two algorithms,the extraction process of the midsagittal plane in the brain is divided into two steps.The first step is to use the global symmetry of the brain edge information to perform a single-scale search on the three-dimensional image to initially locate the mid-sagittal plane.The second step first adjusts the direction of the CT image by using the position of the initial plane to satisfy the requirement of the CT image direction based on the method of extracting the MSP based on the two-dimensional image;Then,using the characteristics that there are local symmetry near the longitudinal crack of the brain in the two-dimensional image,the approximate longitudinal segment is obtained.Removing the outliers in the approximate longitudinal split line segment to obtain an approximate longitudinal split line segment that dominates the mid sagittal plane orientation;finally,the midsagittal plane is fitted.The algorithm not only no longer limits the initial direction of the CT image of the brain,but also ensures the accuracy of the extraction result on the two-dimensional image,and reduces the computational cost.The improved extraction algorithm is compared with the above two methods.The experimental results show that the proposed algorithm can extract the sagittal plane more accurately than the above two traditional methods.