Role of Liver Function Assessment in Portal Venous Interventions and Locoregional Therapies for Liver Tumors

Functional liver assessment has undergone major evolution over the past 30 years. Assessment techniques have progressed from biochemical markers to advanced imaging techniques, particularly nuclear medicine imaging, to improve accuracy. The focus of assessment has also shifted from global hepatic function to precise segmental evaluation. This paradigm shift has greatly influenced liver-directed therapies such as hepatic resection, augmentation, transplantation, and interventional radiology. The aim of this review was to synthesize key developments in liver function assessment, including nuclear medicine techniques, hepatocyte-specific magnetic resonance imaging (MRI), and emerging multimodal and artificial intelligence (AI)-driven approaches. Select literature from the past three decades was examined, with an emphasis on innovations, validation studies, and clinical applications. Incorporating Tc-99 m mebrofenin hepatobiliary scintigraphy (HBS) into liver function assessment allowed quantitative, regional mapping of hepatocellular function. Gadoxetate-enhanced dynamic magnetic resonance imaging (MRI) has created an opportunity for high-contrast resolution images and noninvasive assessment of hepatocyte uptake and excretion. Novel positron emission tomography tracers such as C-11 acetate enabled the combined evaluation of parenchymal function and tumor biology. Studies exploring multimodal fusion, artificial intelligence (AI)-assisted image processing, and incorporating deep learning into predictive models for individualized risk assessment are underway. Overall, integrating anatomical and functional data has become increasingly widespread, allowing for a more precise evaluation of liver function and improvement of therapeutic outcomes. However, reproducibility, standardization, and accessibility remain potential impediments to broad implementation. Future directions for research include prospective validation studies, multimodal fusion, utilizing deep learning in risk stratification, and the development of individualized treatment strategies tailored to patient-specific risk profiles.