| Title: | Video based deep learning for Cardiac MRI |
| Lead Institution: | Stanford University |
| Principal investigator: | Dr William Hiesinger |
Application 71226
WARNING: the interactive features of this website use CSS3, which your browser does not support. To use the full features of this website, please update your browser.
About
Deep learning and artificial intelligence have the potential to transform healthcare delivery. In the past few years, deep learning has shown expert level performance in detecting abnormalities in a variety of medical imaging methods. Aims: In this project we aim to develop video-based deep learning models that learn from magnetic resonance imaging (MRI) scans of the heart. Specifically, we will first train our deep learning models to automatically take measurements of the heart. The proposed project is expected to take 9 months to complete, and we plan to make our codebase and trained models publicly accessible for other researchers to build off our work. Scientific rationale: Previous work on cardiac imaging has shown that it is possible to automate taking measurements from the heart. More importantly however, we now find that medical images contain more 'hidden data' than was previously thought. It is possible for example, to identify cardiovascular risk factors just by looking at an individuals retinal scans. We aim to develop artificial intelligence methods that are capable of 1. quantifying measures of heart function, 2. identifying structural abnormalities, 3. detecting features from the images that are not immediately obvious to clinicians (high blood pressure for example, or diabetes). Project duration: 9 months Public Health impact: The creation of deep learning tools that are open sourced for the greater research community will advance the science of heart disease and the applications of artificial intelligence on cardiac MRI and cardiovascular health in general.2 Publications
| Pub ID | Title | Author(s) | Year | Journal |
|---|---|---|---|---|
| 18197 | A generalizable deep learning system for cardiac MRI | Rohan Shad (+18) | 2026 | Nature Biomedical Engineering |
| 18486 | Sex disparities in deep learning estimation of ejection fraction from cardiac magnetic resonance imaging | Dhamanpreet Kaur (+8) | 2026 | npj Digital Medicine |
Enabling scientific discoveries that improve human health