Deep learning-based system to predict hepatocellular carcinoma resection volume using contrast-enhanced CT.
Authors
Affiliations (8)
Affiliations (8)
- Department of Hepatobiliary Surgery, Chinese PLA 970th Hospital, 7 Zhichu South Road, Zhifu District, Yantai, 264001, People's Republic of China.
- Medical Big Data Research Center, Chinese PLA General Hospital, 28 Fuxing Road, Haidian District, Beijing, 100853, People's Republic of China.
- Medical School of Chinese PLA, 28 Fuxing Road, Haidian District, Beijing, 100853, People's Republic of China.
- China National Clinical Research Center for Neurological Diseases, 119 Fan Yang Road, Fengtai District, Beijing, 100071, People's Republic of China.
- School of Medicine, Nankai University, 94 Weijin Road, Nankai District, Tianjin, 300071, People's Republic of China.
- Department of Hepatobiliary Surgery, Affiliated Yantai Yuhuangding Hospital of Qingdao University, 20 Yuhuangding East Road, Zhifu District, Yantai, 264001, People's Republic of China.
- Department of Hepatobiliary Surgery, Chinese PLA 970th Hospital, 7 Zhichu South Road, Zhifu District, Yantai, 264001, People's Republic of China. [email protected].
- Medical Big Data Research Center, Chinese PLA General Hospital, 28 Fuxing Road, Haidian District, Beijing, 100853, People's Republic of China. [email protected].
Abstract
Accurate calculation of liver resection volume from contrast-enhanced computed tomography images is an essential process in preoperative surgical planning for precise radical resection, which is labor-intensive, time-consuming, and prone to inter-observer variability. We propose a system based on artificial intelligence, named Liver Resection Volume Calculation with Deep learning, that can calculate the planned liver resection volume with high accuracy and efficiency. Our system is trained and tested on medical imaging scans of 990 pathology-confirmed hepatocellular carcinoma patients from two tertiary hospitals collected between January 2012 and December 2022. This system reduces the time consumed by nearly twenty folds and achieves comparable results in calculating parenchymal hepatic resection rate for variable surgery types compared to the manual process based on 3D simulation software used in the hospital currently. Its consistency with the planning results of experienced surgeons demonstrates its applicability in clinical workflow.