Machine Learning Model Using Pre-Cancer Therapy Cardiac Magnetic Resonance Images to Predict Cancer Therapy-Related
Christopher Yu1, Mohammad Peikari2, Dina Labib3
1Department of Medicine, Division of Cardiology, Ted Rogers Program in Cardiotoxicity Prevention, Peter Munk Cardiac Centre, Toronto General Hospital, University Health Network, University of Toronto, Toronto, Ontario, Canada.
JACC. Cardiovascular Imaging
|May 29, 2026
Summary
Deep learning models using cardiac MRI before cancer therapy can predict cardiac dysfunction risk. These AI models outperform traditional clinical and imaging methods for early detection.
Area of Science:
- Cardiology
- Oncology
- Artificial Intelligence
- Medical Imaging
Background:
- Predicting cancer therapy-related cardiac dysfunction (CTRCD) is a significant clinical challenge.
- Early identification of patients at risk is crucial for timely intervention and management.
Purpose of the Study:
- To evaluate deep learning (DL) models using pre-treatment cardiac magnetic resonance (CMR) images for predicting CTRCD.
- To compare the predictive performance of DL models against conventional clinical and imaging-based risk scores.
Main Methods:
- Utilized CMR short-axis cine images from HER2+ breast cancer patients undergoing anthracycline and trastuzumab therapy.
- Developed deep convolutional neural network architectures for image-based DL models to predict CTRCD.
- Validated models internally and externally using data from prospective studies.
Main Results:
- The DL model achieved an AUC of 0.85 and F1 score of 0.69 in internal validation for CTRCD prediction.
- External validation demonstrated an AUC of 0.80 and F1 score of 0.55 for the DL model.
- DL models significantly outperformed clinical risk scores (AUC 0.66) and conventional CMR/echocardiography parameters (AUCs 0.59-0.62).
Conclusions:
- DL models leveraging pre-therapy CMR images offer superior prediction of future CTRCD risk.
- This AI-driven approach surpasses traditional methods in identifying patients susceptible to cardiotoxicity.
- Pre-treatment CMR analysis with DL holds promise for personalized cancer treatment strategies.
