BCLC Classification and AI-Based Image Quantification in HCC: Journal of Hepatology | March 2026
Introduction
The Barcelona Clinic Liver Cancer (BCLC) classification has remained the cornerstone for staging, prognosis, and treatment allocation in hepatocellular carcinoma for more than two decades. Its strength lies in its simplicity and clinical applicability, integrating tumor burden, liver function, and performance status into a unified framework. However, modern imaging contains far more quantitative information than is currently utilized in BCLC. With the rapid evolution of artificial intelligence, there is now an opportunity to extract complex imaging features automatically and integrate them into clinical decision-making. This article explores how AI can complement, rather than replace, the BCLC system to enable more precise and personalized management of HCC.
Summary
This review highlights that while BCLC remains robust and widely validated, it underutilizes the wealth of data embedded in imaging. AI-based tools can quantify tumor volume, assess intratumoral heterogeneity, detect vascular invasion, evaluate portal hypertension through surrogate markers such as spleen volume, and even estimate body composition to reflect patient performance status. These parameters have shown strong associations with prognosis and treatment response in early studies. However, despite promising results, most AI applications remain confined to retrospective research settings due to poor reproducibility, lack of standardization, absence of prospective validation, and limited integration into clinical workflows. The article emphasizes that the key challenge is not technological capability but translational implementation. AI must be seamlessly embedded into routine workflows, supported by structured reporting, validated across diverse populations, and aligned with clinical frameworks such as BCLC. Importantly, the authors reinforce that clinical decision-making in HCC is inherently influenced by complexity, uncertainty, subjectivity, and emotion, and AI should support, not replace, this human-centered process. The future likely lies in a hybrid model where BCLC provides the clinical backbone, and AI adds quantitative refinement to improve prognostication and treatment selection.
