Development of an Interactive Semantic Medical Image Segmentation System

Medical image segmentation (i.e., liver segmentation) is an important pre-processing for computer-aided diagnosis and computer-aided surgery. Deep learning, which can perform fully automated segmentation, has been popular in the medical field because of its ability to reduce the burden on doctors. However, segmentation is inherently a difficult problem and the accuracy of current deep learning-based segmentation methods remains insufficient. In this work, an interactive network-based segmentation is proposed, which can improve the operation through collaborations with a doctor.