Performance of Artificial Intelligence Models in Predicting Responsiveness of Hepatocellular Carcinoma to Transarterial Chemoembolization (TACE): A Systematic Review and Meta-Analysis.

Abstract

Background: Hepatocellular carcinoma (HCC) remains a leading cause of cancer-related mortality worldwide. Handcrafted radiomics (HCR) and deep learning (DL) models have emerged as promising predictive tools extracting granular insights from images.

Objective: This systematic review and meta-analysis aims to evaluate the predictive performance of these artificial intelligence models in predicting treatment efficacy in patients with HCC who are undergoing transarterial chemoembolization (TACE).

Methods: A comprehensive search was conducted on PubMed, Embase, Web of Science, and Cochrane Library databases up to June 15, 2024. Inclusion criteria encompassed studies involving patients with confirmed HCC undergoing TACE. Random-effects diagnostic test accuracy meta-analyses were performed using bivariate modeling. Methodological quality was assessed using the Prediction model Risk of Bias Assessment Tool.

Results: Twenty-seven studies were included in this review. The overall meta-analysis of models for TACE treatment response included 11 studies. The pooled area under the receiver operating characteristic curve was 0.89 (95% confidence interval: 0.81-0.93) for internal validation and 0.81 (95% confidence interval: 0.80-0.92) for external validation, with no significant differences (P = .66). Moreover, no significant differences were found between DL and HCR models (P = .21) or between models with and without clinical data (P = .78).

Conclusion: Artificial intelligence models, including DL and HCR, show potential for predicting treatment outcomes in patients with HCC who are undergoing TACE. However, the observed heterogeneity stresses the need for further research.