AI-Generated 3D Liver Models Predict Surgical Complexity Better Than Surgeons: Annals of Surgery | October 2026

What Did the Study Find?

This study developed an automated artificial intelligence system that converts preoperative CT scans into patient-specific 3D liver models to predict the technical complexity of liver resection.

• The AI system automatically reconstructed the liver, tumours and major vascular structures using deep learning.
• Researchers introduced a novel anatomical framework called the Hepatic Central Zone (HCZ) to quantify tumour proximity to critical hepatic vessels.
• HCZ reconstruction achieved a Dice similarity score of 82% ± 4.6%, including in patients with atypical vascular anatomy.
• The automated pipeline was evaluated in 145 patients with hepatocellular carcinoma undergoing liver resection.
• AI predicted intraoperative liver resection complexity with 79.4% ± 3.4% accuracy.
• The model achieved an AUC of 0.851 ± 0.032, outperforming individual surgeons and their combined preoperative assessments based on CT imaging.
• Unlike traditional major/minor hepatectomy classifications, the model incorporated patient-specific tumour location and vascular relationships, rather than simply the number of segments removed.

Why Does It Matter?

Conventional surgical complexity classifications often fail to capture the anatomical difficulty of operating near major hepatic vessels.

AI-generated virtual liver models could provide a more objective assessment of operative complexity, potentially improving surgical planning, case allocation and referral to specialised hepatobiliary centres.

GastroAGI Takeaway

The future of liver surgery may begin with an AI-generated digital twin—before the surgeon enters the operating room.

By automatically mapping the liver, tumour and vascular anatomy, this AI system predicted operative complexity more accurately than surgeons assessing conventional CT scans.

Its greatest potential may be in identifying technically challenging resections early and directing patients to appropriately experienced surgical teams.