Liver Segmentation Based on Convolutional Neural Network and Superpixel in CT Image
JIANG Tao1, LIN ChungChih2, WU Shuicai1, WANG Xiaoru1, LIN Yuanping2
1. College of Life Science and Bioengineering, Beijing University of Technology, Beijing 100124, China;
2. Department of Computer Science and Information Engineering, Chang Gung University, Taoyuan Taiwan 33302, China
Abstract:Liver cancer is one of the major diseases that threaten human health. The accurate segmentation of liver tissue from
medical imaging is an important part of computer-aided diagnosis and surgical planning of liver diseases. Due to individual
differences in the liver, similar gray values of surrounding organs, and other factors, it is difficult to accurately segment the liver
from CT images. An automatic liver segmentation method based on convolutional neural network and superpixel from CT image is
proposed. Firstly, the convolutional neural network is used for target detection, and the liver region is automatically located. Then
the liver is segmented by the superpixel algorithm. Finally, corrosion, expansion, median filtering and other post-processing are
performed. In this paper, the proposed automatic liver segmentation algorithm was evaluated and verified by the 3DIRCADb public
dataset. The results showed that the DICE index of the automatic segmentation of the liver was 0.951, the VOE index was 0.0917,
and the RVD index was -0.018, which was showed a good segmentation accuracy.
姜涛1,林仲志2,吴水才1,王笑茹1,林沅平2. 基于卷积神经网络和超像素的CT图像肝脏分割[J]. 中国医疗设备, 2020, 35(2): 72-76.
JIANG Tao1, LIN ChungChih2, WU Shuicai1, WANG Xiaoru1, LIN Yuanping2. Liver Segmentation Based on Convolutional Neural Network and Superpixel in CT Image. China Medical Devices, 2020, 35(2): 72-76.