Machine learning model predicts hepatocellular carcinoma treatment response

Aug. 18, 2022 — Machine learning models applied to currently underutilized imaging features could help build more reliable organ allocation and liver transplant eligibility criteria, according to ARRS’ American Journal of Radiology (AJR).

“The findings demonstrate that a machine learning-based model can predict recurrence prior to treatment allocation in patients with early-stage hepatocellular carcinoma (HCC) initially eligible for liver transplantation,” wrote corresponding author Julius Chapiro of Yale’s Department of Radiology and Biomedical Imaging. New Haven School of Medicine, Connecticut.

The proof-of-concept study by Chapiro and colleagues included 120 patients (88 men, 32 women; median age, 60 years) with early-stage HCC diagnosed between June 2005 and March 2018 who were initially eligible for liver transplantation condition and received transplantation, resection, or thermal ablation. Patients underwent pre-treatment MRI and post-treatment imaging monitoring, and image features were extracted from the post-contrast phase of pre-treatment MRI using a pretrained convolutional neural network (VGG-16). Pretreatment clinical features (including laboratory data) and extracted imaging features were integrated to develop three ML models—clinical, imaging, combined—for recurrence prediction 1-6 years after treatment.

Ultimately, all three models predicted post-treatment recurrence of early-stage HCC from pre-treatment clinical (AUC 0.60-0.78 for all six time frames), MRI (AUC 0.71-0.85), and both data combined (AUC 0.62-0.86). Using imaging data as the sole model input yielded higher predictive performance than using clinical data alone; however, combining the two data types did not significantly improve performance compared to using imaging data alone.

For more information:

Leave a Comment

Your email address will not be published.