New AI Creates Heart Strain Maps Without Extra Scans

Heart disease is a leading cause of death, and doctors often use MRI scans to check how well the heart is pumping. A key measure is 'strain'—how much the heart muscle stretches and squeezes with each beat. Traditionally, getting accurate strain maps requires a special MRI technique called DENSE, which takes extra time and equipment. Now, researchers have created an AI tool called GenStrain that can produce high-quality strain maps from standard heart MRI videos, which are already taken during routine check-ups.

GenStrain works by learning from the detailed motion patterns captured by DENSE scans. It uses a diffusion model, a type of AI that generates data by gradually refining random noise into realistic patterns. In this case, it generates motion maps that are as accurate as those from DENSE, but using only the standard MRI images. The AI also incorporates information about the heart's structure and motion from a separate model to guide the generation. In tests, GenStrain outperformed existing methods, including traditional and deep learning approaches.

This matters because it could make advanced heart assessments more accessible. Patients wouldn't need extra scans, which means shorter exam times, lower costs, and less discomfort. Doctors could get detailed strain information from routine MRIs, potentially leading to earlier detection of heart issues. The code is publicly available, so other researchers and clinicians can build on this work. While more testing is needed before it's used in hospitals, GenStrain represents a promising step toward smarter, more efficient heart care.