AI in steatohepatitis diagnostics: precision beyond the microscope

Non-alcoholic steatohepatitis (NASH), the progressive form of nonalcoholic fatty liver disease (NAFLD), is a major global health burden without curative therapy. Conventional histological diagnosis is limited by interobserver variability and insensitivity to subtle changes, necessitating more objective diagnostic tools. Artificial intelligence (AI), integrated with digital pathology techniques such as second harmonic generation/two-photon excitation (SHG/TPE) fluorescence imaging, provides quantitative and reproducible assessment of steatosis, ballooning, and fibrosis. Recent studies have demonstrated strong correlations between AI-derived and pathologist-assigned grades, underscoring AI's potential to enhance diagnostic precision and reproducibility. However, challenges including high operational costs, limited large-scale validation, population heterogeneity, and lack of regulatory frameworks remain barriers to clinical translation. Expanding research, developing standardized protocols, and establishing robust policies are essential for widespread adoption. AI-based pathology could revolutionize the diagnostic paradigm for liver disease, enabling earlier detection, improved patient stratification, and precision care in NASH.