Journal: Hepatology (Baltimore, Md.)
This publication is a narrative review on the role of artificial intelligence (AI) in the management of primary liver cancer, focusing on hepatocellular carcinoma and intrahepatic cholangiocarcinoma.
Key points:
- Clinical problem: Primary liver cancer remains a major cause of cancer death. Curative options depend on early-stage detection, but current surveillance (ultrasound) and diagnostic imaging (contrast-enhanced CT/MRI) are limited by operator dependence, suboptimal sensitivity for small lesions, and variability in interpretation.
- AI in surveillance and imaging:
- • Deep learning models can enhance ultrasound by improving detection of small tumors, enabling automated triage, and potentially reducing radiologist workload.
- • On CT/MRI, AI systems can reach expert-level performance in lesion detection, segmentation, and characterization.
- • Integration with standardized reporting systems (e.g., LI-RADS-type frameworks) may improve consistency and reproducibility of imaging interpretation.
- AI in pathology:
- • Digital pathology algorithms can differentiate hepatocellular carcinoma from intrahepatic cholangiocarcinoma.
- • AI can classify dysplastic nodules and may predict future cancer development from histologic slides, supporting risk stratification.
- Emerging directions:
- • Foundation models and multimodal AI are highlighted as a next step, aiming to integrate radiologic, pathologic, and molecular data for comprehensive, patient-specific disease modeling.
- Barriers to implementation:
- • Major challenges include data privacy and sharing, regulatory approval pathways, cost and resource requirements, and risks of algorithmic bias and inequity.
- • The authors emphasize the need for large, prospective, multicenter validation studies to demonstrate real-world clinical benefit and safety.
- Overall conclusion:
- • Carefully validated, explainable, and trustworthy AI tools have the potential to enable earlier detection, improve diagnostic accuracy, standardize care, and support more equitable outcomes in liver cancer, but robust clinical validation and thoughtful integration into practice are essential.