基于C-V模型的肝脏病灶CT图像多相分割

Multiphase Segmentation of Liver Lesion in CT Images Based on C-V Model

目的:研究一种半自动的适用于CT图像中肝脏病变区域比较复杂的图像分割方法。方法:C-V模型和多相水平集相结合的分割方法,C-V模型依靠图像的全局性质使演化曲线最终停留在目标物体的边界,达到分割目标物的目的。多相水平集方法的引入,使得分割多目标区域成为可能,同时还避免了水平集函数过多而带来的覆盖区域的重叠和真空问题。结果:经过多次迭代,肝脏病变区域的低密度区的癌变区域和高密度区都被很好地分割了出来,实现了分割肝脏复杂区域的目的。结论:各目标区域都能够很好地分割出来,效果较好。

Objective To study a semi-automatic segmentation framework for liver complex lesion in CT images. Methods A multiphase level set method of image segmentation based on C-V model was proposed, which was in connection with the complex information in the pathological changes area of liver. Depending on the overall characteristics of the image, the active curve contour of C-V model stopped on the edge of objects. It was possible for the segmentation of multi-objects because multi-phase level set was drawn into. Simultaneously, the problem of overlapping and hollow caused by more level set function was avoided. Results The method behaved well in the segmentation of CT images of the liver lesion, two different area were separated. Conclusion The segmentation tests prove that the proposed segmenting method makes a good result.