Fuzzy C-means clustering-based multilayer perceptron neural network for liver CT images automatic segmentation
Yuqian Zhao, Yunlong Zan, Xiaofang Wang, Guiyuan Li · 2010
A new liver segmentation algorithm is proposed. First, the threshold method was used to remove the ribs and spines in the initial image, and the fuzzy C-means clustering algorithm and morphological reconstruction filtering were used to segment the initial liver CT image. Then the multilayer perceptron neural network was trained by the segmentation result of initial image with the back-propagation algorithm. The adjacent slice CT image was segmented with the trained multilayer perceptron neural network. Last, morphological reconstruction filtering was used to smooth the contour of the liver edge. The experimental results show that the proposed algorithm can effectively segment the livers from CT images, despite the gray level similarity of adjacent organs and different gray level of tumors in the liver.