4D MS lesion segmentation method based on temporal information
1.School of Computer Science and Communication Engineering, Jiangsu University, Zhenjiang, Jiangsu 212013, China; 2.North Information Control Group Co., Ltd., Nanjing, Jiangsu 211153, China
Abstract:To solve the problem that 3D brain magnetic resonance (MR) image segmentation method could not accurately segment 4D multiple sclerosis (MS) lesion, integrating the information of each time point in 4D data, a 4D MS lesion segmentation method was proposed by building a dynamic prior information from adjacent time point data to improve the level set method driven by the local information. The 4D MS lesion segmentation algorithm was obtained by solving the minimum of the proposed model based on the alternative iteration algorithm and the steepest descent method. To validate the effect of the method, the real clinical 4D data were tested and compared with the results obtained by 3D segmentation method. The results show that the proposed 4D MS lesion segmentation method can introduce the temporal information constraint to obtain more accurate segmentation results and realize efficiently quantitative analysis of the following lesions.