Machine Learning using MR İmaging Radiomics can Predict the Response of Large Hepatocellular Carcinoma to...

Area of Science:

• Radiology
• Oncology
• Artificial Intelligence

Background:

• Large hepatocellular carcinoma (HCC) tumors (>5 cm) are linked to poorer patient outcomes.
• Transarterial radioembolization (TARE) is a treatment for HCC, but predicting response in large tumors remains challenging.
• Radiomics and machine learning (ML) have not been extensively used for predicting TARE response in large HCC lesions.

Purpose of the Study:

• To evaluate the effectiveness of ML models incorporating radiomics features in predicting TARE treatment response for large HCC lesions.
• To compare the performance of radiomics-based ML models against traditional clinical models.

Main Methods:

• The study analyzed 49 patients with large HCC (>5 cm) who underwent TARE.
• Treatment response was assessed using modified Response Evaluation Criteria in Solid Tumors (mRECIST) at 3-month follow-up MRI.
• Radiomics features were extracted from pre-treatment contrast-enhanced T1-weighted (CE-T1) and T2-weighted (T2W) MRI images to train classification ML models.

Main Results:

• Radiomics models using CE-T1 and T2W images achieved an accuracy of 79.6% with AUCs of 0.92 and 0.77, respectively.
• A clinical model showed 77.6% accuracy and an AUC of 0.65.
• No statistically significant difference was observed between the radiomics and clinical models (P = 0.092).

Conclusions:

• Machine learning models utilizing pre-treatment radiomics features from MRI show potential for predicting treatment response in patients with large HCC lesions undergoing TARE.
• Radiomics analysis may offer valuable insights for tailoring TARE treatment strategies for large HCC.