A Fully Automated Hybrid Approach to Assessing Liver Fibrosis and Necroinflammation on Conventional MRI:A Multi-center Study

To our knowledge, this is the first study developing a multi-task hybrid models incorperating conventional MR tissue texture and routine clinical biomarkers with both good accuracy and explainability in detecting fibrosis and necroinflammation. Our study used an interactive deep learning approach to automatedly segment the entire volumetric liver contours more effectively. Our CoRC models outperformed routine clinical fibrotic scores (FIB-4, APRI), and TE-LSM by discrimination, calibration in the large multicenter cohorts. Our CoRC models could be as a potential alternative when biopsy, hepatobiliary phase (HBP) images, liver stiffness measurement (LSM) are unavailable.