Automated Delineation of Couinaud Segments on CT for Future Liver Remnant Volumetry.
Authors
Affiliations (5)
Affiliations (5)
- Radiology and Imaging Sciences, National Institutes of Health (NIH) Clinical Center, 10 Center Dr, Bldg 10, Rm 1C224, Bethesda, MD 20892.
- Center for Cancer Research, National Cancer Institute (NCI), Bethesda, Md.
- Liver Diseases Branch, National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), Bethesda, Md.
- Laboratory of Pathology, Center for Cancer Research, National Cancer Institute (NCI), Bethesda, Md.
- School of Medicine and Public Health, University of Wisconsin-Madison, Madison, Wis Received October 2, 2025; revision requested XXX; revision received March 30, 2026; accepted.
Abstract
Purpose To develop a deep learning model that automatically delineates the eight liver Couinaud segments and the spleen on CT for future liver remnant (FLR) volumetry. Materials and Methods In this retrospective study (January 2001 and October 2025), eight liver Couinaud segments and the spleen were manually labeled on CT scans of patients from Institution-A and the public Medical Segmentation Decathlon dataset. A 3D nnU-Net segmentation model was trained on this dataset and evaluated on three datasets (one internal and two external). Results The training dataset included 498 patients (442 from the public Medical Segmentation Decathlon dataset and 56 from Institution-A, mean age 55 ± 7 [SD] years, 38 males), while the testing dataset included 64 patients from Institution-A (50 had liver fibrosis and 8 underwent portal vein embolization; PVE), 197 patients from the publicly available colorectal liver metastases (CRLM) dataset (mean age 59 ± 12 years, 117 males), and 50 patients (25 were healthy and 25 had cirrhosis) from an external Institution-B (mean age 49 ± 9 years, 29 males). For the whole liver in Institution-A and Institution-B, Dice scores of 0.98 ± 0.02 (95% CI: 0.97, 0.99) and 0.98 ± 0.03 (95% CI: 0.97, 0.99), and 95% percentile Hausdorff Distance (HD) errors of 2.5 ± 3.8 mm (95% CI: 1.6, 3.3) and 3.3 ± 6.6 mm (95% CI: 1.4, 5.2) were obtained, respectively. The pre-PVE <i>FLR</i><sub>%</sub> and post-PVE <i>FLR</i><sub>%</sub> volume differences (manual vs automated, 8 patients) were 0.03 ± 2.4 and-0.39 ± 3.0, respectively. For the FLR in the CRLM dataset, a Dice score of 0.99 ± 0.01 (95% CI: 0.99, 0.993) and an HD error of 0.9 ± 1.8 mm (95% CI: 0.6, 1.1) were achieved. Conclusion The model accurately estimated preoperative FLR volumetry and generalized well to patients with colorectal liver metastases, fibrosis, cirrhosis and healthy controls. ©RSNA, 2026.