Interactive Liver Segmentation in CT Volumes Using Fully Convolutional Networks

Abstract

Organ segmentation is one of the most fundamental and challenging task in computer aided diagnosis (CAD) systems, and segmenting liver from 3D medical data becomes one of the hot research topics in medical analysis field. Graph cut algorithms have been successfully applied to medical image segmentation of different organs for 3D volume data which not only leads to very large-scale graph due to the same node number as voxel number. Slice by Slice liver segmentation method is one of the technique that normally used to solve the memory usage. However, the computation times are increased and reduce the accuracy. In this paper we propose an interactive organ segmentation using fully convolutional networks. The network will perform slice by slice which only 1 slice of seed points in each volume. To validate effectiveness and efficiency of our proposed method, we conduct experiments on 20 CT volumes, focus on liver organ and most of which have tumors inside of the liver, and abnormal deformed shape of liver. Our method can segment with 0.95401 dice accuracy with better than stage-of-the-art methods.