Radiomics Analysis of Preprocedural CT Imaging for Outcome Prediction after Transjugular Intrahepatic Portosystemic...

Area of Science:

• Medical Imaging and Radiomics
• Interventional Radiology
• Machine Learning in Healthcare

Background:

• Cirrhosis management often involves transjugular intrahepatic portosystemic shunt (TIPS) procedures.
• Predicting patient outcomes after TIPS is crucial for treatment planning and resource allocation.
• Radiomics offers a non-invasive method to extract quantitative features from medical images for predictive modeling.

Purpose of the Study:

• To evaluate the efficacy of radiomics in predicting preoperative outcomes in cirrhotic patients undergoing TIPS.
• To assess the role of radiomics in forecasting survival and clinical response following TIPS with controlled expansion covered stents.

Main Methods:

• Retrospective analysis of preoperative CT scans from 76 cirrhotic patients who underwent TIPS.
• Segmentation of the whole liver into Volumes of Interest (VOIs) during unenhanced and portal venous phases.
• Extraction and analysis of radiomics features, with outcome prediction assessed using receiver operating characteristic (ROC) curves.

Main Results:

• Radiomics models demonstrated the highest predictive performance for 6-month overall survival (AUROC 0.767) and clinical response (AUROC 0.755).
• At the portal venous phase, sensitivity and accuracy for outcome prediction were 74.00% and 65.34%, respectively.
• The study identified specific radiomics features with potential to predict patient outcomes post-TIPS.

Conclusions:

• A machine learning-based CT radiomics algorithm applied pre-interventionally can aid in predicting survival and clinical response after TIPS in cirrhotic patients.
• Radiomics shows potential as a valuable tool for personalized treatment strategies in patients undergoing TIPS.