Automatic Aortic Dissection Recognition Based on CT Images
Xiaojie Duan, Xiaobing Shi, Jianming Wang, Qingliang Chen · 2018
In order to improve diagnostic rate of aortic dissection, the aortic dissection three-dimensional (3D) reconstruction system is indispensable. Recognition and segmentation of aortic dissection is an indispensable part of 3D reconstruction. Thus, an automatic recognition algorithm of aortic dissection based on CT images is proposed in this paper. According the characteristics of CT image noise, we choose filtering and sharpening to preprocess the image. The GVF Snake algorithm is used when segment aortic dissection in CT images, and then the edge information of the aortic region is extracted by the Sobel operator. The edge image of the aorta is projected by the Radon transformation principle, and the projection data is combined with the perimeter and area characteristics of the aortic region to judge whether the aortic dissection exists. In order to reduce the harm caused by misjudgment, the minimum Bayes risk decision can be used.