Liver Segmentation in CT Images with Adversarial Learning

Yuxuan Chen, Suiyi Li, Su Yang, Wuyang Luo

Abstract: Liver diseases, especially liver cancer, are a major threat to human health. In order to assist doctors in efficiently diagnosing the condition and developing treatment plans, automatic segmentation of the liver from CT images has a strong clinical need. However, it is difficult to design an accurate segmentation algorithm because of the blurred boundary of CT images and the great difference of pathological changes. In this paper, we propose a cascade model for liver segmentation from CT images, which uses cascade U-nets with adversarial learning to obtain more accurate segmentation results. The experimental results show that the proposed algorithm is competitive and its dice value reaches 0.955.