Abstract: Hepatocellular carcinoma (HCC) immunotherapy is significantly constrained by a low objective response rate (~30%) and the lack of universally applicable predictive biomarkers. Radiomics, a non-invasive technique that extracts high-throughput hidden features from medical images, offers innovative solutions for patient selection, treatment response assessment, and prognosis prediction. This review systematically summarizes the application scenarios, feature extraction, and model construction of multimodal imaging data in HCC immunotherapy. It highlights advances in radiomics for predicting treatment response and evaluating the tumor immune microenvironment (TIME) and underlying molecular signatures. It also analyzes key challenges, including limited sample sizes and poor model generalization, and outlines future directions such as multicenter standardized studies and multi-omics integration. The goal is to inform the clinical translation of radiomics for precision management of HCC immunotherapy. Keywords: hepatocellular carcinoma, immunotherapy, radiomics, prognosis prediction, TIME, deep learning
Wang et al. (Fri,) studied this question.