MRI-Based Deep Learning and Radiomics Nomogram for Predicting Hepatocellular Carcinoma Recurrence Within Six Months...

Area of Science:

• Medical Imaging
• Oncology
• Artificial Intelligence

Background:

• Hepatocellular carcinoma (HCC) recurrence after thermal ablation is a clinical challenge.
• Accurate prediction of early recurrence is crucial for timely intervention and improved patient outcomes.
• Existing prediction models may not fully leverage advanced imaging and clinical data.

Purpose of the Study:

• To develop and validate a magnetic resonance imaging (MRI)-based deep learning (DL)-radiomics (Rad)-clinical nomogram.
• To predict early recurrence of HCC within six months following thermal ablation.
• To enhance the accuracy of risk stratification for HCC patients.

Main Methods:

• Retrospective analysis of BCLC stage 0-A HCC patients who underwent dynamic contrast-enhanced MRI.
• Development of clinical, DL score, and Rad score models using logistic regression and feature selection techniques.
• Construction of a combined DL-Rad-Clinical nomogram integrating imaging features, DL scores, and clinical factors (LnAFP, low signal lesions).

Main Results:

• The DL-Rad-Clinical nomogram demonstrated strong predictive performance in the training set (AUC = 0.896).
• In the test set, the nomogram showed a higher AUC (0.774) compared to other models, indicating its potential clinical utility.
• Key predictive factors included Rad score, DL score, natural logarithm of alpha-fetoprotein (LnAFP), and multiple low signal lesions.

Conclusions:

• The developed DL-Rad-Clinical nomogram effectively identifies HCC patients at risk for early recurrence within six months post-thermal ablation.
• This integrated approach combining DL, radiomics, and clinical data offers a promising tool for personalized risk assessment in HCC management.
• Further validation is warranted to solidify its role in clinical decision-making.