A two-stage approach for automatic liver segmentation with Faster R-CNN and DeepLab

• Published in: Neural computing & applications 2020-06, Vol.32 (11), p.6769-6778
• Main Authors: Tang, Wei, Zou, Dongsheng, Yang, Su, Shi, Jing, Dan, Jingpei, Song, Guowu
• Format: Article
• Language: English
• Subjects: Artificial Intelligence; Computational Biology/Bioinformatics; Computational Science and Engineering; Computer Science; Data Mining and Knowledge Discovery; Image detection; Image Processing and Computer Vision; Image segmentation; Liver; Probability and Statistics in Computer Science; Radiation therapy; S.I. : Brain inspired Computing &Machine Learning Applied Research-BISMLARE; Volume measurement
• ISSN: 0941-0643; 1433-3058
• DOI: 10.1007/s00521-019-04700-0

Proper liver segmentation is a key step in many clinical applications, including computer-assisted diagnosis, radiation therapy and volume measurement. However, liver segmentation is still challenging due to fuzzy boundary, complex liver anatomy, present of pathologies, and diversified shape. This paper presents a novel two-stage liver detection and segmentation model DSL. The first stage uses improved Faster Regions with CNN features (Faster R-CNN) to detect approximate position of liver. The obtained images are processed and input into DeepLab to obtain the contour of liver. The proposed approach is validated on two datasets MICCAI-Sliver07 and 3Dircadb. Experimental results reveal that the proposed method outperforms the state-of-the-art solutions in terms of volume overlap error, average surface distance, relative volume difference, and total score.