Dual-Phase Computed Tomography-Based Deep Learning Architecture for Three-Year Survival Prediction in Hepatocellular...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Oncology
• Hepatocellular Carcinoma Research

Background:

• Hepatocellular carcinoma (HCC) presents a significant global health challenge, being the sixth most common cancer and third leading cause of cancer mortality.
• Computed tomography (CT) is crucial for HCC evaluation, but interpretation can be complex and variable.
• Deep learning offers advancements in computer-aided diagnosis (CADx) to improve efficiency and accuracy in medical image analysis.

Purpose of the Study:

• To develop and evaluate a deep learning framework for predicting three-year survival in HCC patients using dual-phase CT.
• To enhance prognostic modeling by integrating phase-specific imaging features and clinical variables.

Main Methods:

• A dual-branch deep learning model (MedNeXt) was designed to process arterial and venous phase CT images.
• A Dual Phase Contextual Fusion Block (DPCFB) was used for cross-phase feature integration.
• A Cascaded Damper Block (CDB) incorporated clinical data and tumor size for improved prognostic accuracy.

Main Results:

• The proposed deep learning system achieved 85% accuracy, 83% sensitivity, and 86% specificity.
• The Area Under the ROC Curve (AUC) reached 0.88, indicating strong predictive performance.
• The model effectively integrated multiphase CT data with clinical information for survival prediction.

Conclusions:

• Combining multiphase CT imaging with clinical data using deep learning enables accurate prediction of long-term survival in HCC patients.
• The developed framework shows potential as a clinical decision-support tool for HCC prognosis, treatment planning, and patient management.