The Potential of Artificial Intelligence to Improve Selection Criteria for Liver Transplantation in HCC

Despite improved therapeutic concepts, the survival of patients with hepatocellular carcinoma (HCC) is limited. Liver transplantation (LT) is the best possible treatment for suitable patients. This therapy is of particular importance, because it not only removes the cancer but also cures the underlying structural liver disease. Due to the persistent lack of donor organs, however, the oncological prognosis after LT is of particular importance for fair organ allocation. Bonus points on the organ waiting list are rewarded for tumors within a certain tumor extent. In general, macrovascular invasion and extrahepatic tumor manifestation are considered to be contraindications for LT, as survival in these patients is very low. In recent years, however, microvascular invasion and poorly differentiated tumors have also turned out to be unfavorable. Most selection criteria for LT in HCC are still based on very simple imaging criteria like size and number without utilizing additional imaging characteristics inherent to the tumor nodule, which could be processed in a “virtual biopsy”. Recently, diagnostic research has presented the clinical benefit of artificial intelligence (AI) in the use of deep-learning strategies for digital diagnosis of poorly differentiated or microvascular-infiltrated tumors. In addition, evaluation of TACE response is analyzed as a possibility to estimate LT survival. The aim of this review is to provide an overview of recent advances in HCC diagnosis and to classify the clinical relevance of these diagnostic and technical advances. Secondly, we discuss how these advances could affect the organ allocation process.