Incorporating artificial intelligence into imaging for surveillance and diagnosis of liver cancer: Innovations, challenges, and clinical translation.
Authors
Affiliations (5)
Affiliations (5)
- Department of radiology, Hôpital Beaujon AP-HP Nord, Clichy, France.
- Université Paris-Cité, Inserm, Centre de recherche sur l'inflammation, UMR 1149, Paris, France.
- Department of Pathology, Henri-Mondor Hospital, Assistance Publique-Hôpitaux Paris (AP-HP), Créteil, France.
- Université Paris-Est Créteil (UPEC), INSERM, U955, Créteil, France.
- MINT-Hep, Mondor Integrative Hepatology, Créteil, France.
Abstract
Primary liver cancer, mainly hepatocellular carcinoma and intrahepatic cholangiocarcinoma, remains a leading cause of cancer mortality worldwide. Early detection is crucial for curative treatment, yet current surveillance and diagnostic strategies, primarily ultrasound surveillance and contrast-enhanced CT or MRI, suffer from operator dependence, limited sensitivity, and interpretive variability. Artificial intelligence (AI) offers transformative potential across the liver cancer continuum, from surveillance to diagnosis and pathology. Deep learning-based models have improved ultrasound detection of small liver tumors, enabling automated triage and reducing workload. On CT and MRI, AI systems achieve expert-level performance for lesion detection, segmentation, and characterization, supporting standardized interpretation through frameworks such as LI-RADS. In digital pathology, AI algorithms can distinguish between hepatocellular carcinoma and cholangiocarcinoma, classify dysplastic nodules, and even predict future cancer development from biopsy slides. Recent advances in foundation models and multimodal AI promise to unify radiology, pathology, and molecular data, paving the way for comprehensive, patient-specific disease modeling. However, widespread clinical integration faces major challenges, including data privacy, regulatory approval, cost sustainability, and algorithmic bias. Large, prospective multicenter validation studies are essential to confirm clinical benefit and safety. Ultimately, the careful implementation of trustworthy and explainable AI tools could enable earlier detection, greater diagnostic precision, and more equitable liver cancer care.