Active contours driven by localizing region and edge-based intensity fitting energy with application to segmentation of the left ventricle in cardiac CT images.
Neurocomputing(2015)
摘要
Dual source CT of the heart is a well-known and accepted method for detection of cardiac disease. However, weak edges, touching characters, intensity inhomogeneities and complex background lead LV segmentation to leakage and false boundary, in cardiac CT images. This difficult task is accomplished in this work by establishing a new active contour model in a variational level set formulation. Its external energy functional incorporates an edge-based information fitting term, which is an adaptive diffusion flow (ADF), and responsible for extracting object boundaries, especially segmenting the weak and missing borders, and a localizing region intensity fitting, which localizes the Chan-Vese external energy against intensity inhomogeneity and complex background to improve the robustness of the proposed method. For improving the adaptability of the model, the weighted parameter between them is then designed according to the gradient information of image. Besides, the regularity of the level set function is intrinsically preserved by the length regularization term to ensure the curve smooth. Experimental results demonstrate desirable performance of our extension method for real-world dual source cardiac CT images.
更多查看译文
关键词
image segmentation,level set
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络