This review aims to outline the current literature status of artificial intelligence (AI) applications in characterising liver lesions. Our aim is also to address the challenges and future prospects of detection, segmentation, and classification.

Liver lesion analysis studies still lag behind in numbers when compared to other organs like lung or brain. Artificial intelligence demonstrates significant potential in streamlining workflow and enhancing liver lesion characterization. Addressing current challenges in AI and hepatic imaging is imperative for advancing patient care.

We analysed the literature data from academic databases (Web of Science, PubMed, Scopus) from 2019 to 2023 for published papers on artificial intelligence applied to liver imaging to evaluate the trends of publications on liver lesion characterisation. We also compared the trends of liver lesions to that of brain and lung lesions. 

Hepatic lesions are frequently encountered in clinical practice and AI technology has shown promising results in processing medical images with the opportunity to streamline and increase radiologist’s performance. These technologies show great potential in moving towards automatic detection and quantification of liver lesions as well as classification tasks acting as a support tool for diagnosis. Our analysis shows that most research papers in AI applications for medical data have been applied to brain and lung images with limited advances for liver lesion characterisation despite an impeding need for more quantitative data and a progression towards personalized patient management in abdominal pathologies. This limitation might be due to the complexity of liver imaging with multiphase acquisition and more susceptibility for breathing movement artefacts. Nevertheless, there are studies that show that using quantitative AI markers could improve patient management and would enhance the characterisation of liver lesions.