Curie Brief
Turn on cookies to sign in
Signing in saves your progress to your Curie account. We can only do that with cookies on — turn them on to continue.
News
AI reads your heart scan like never before. A new analysis from the SCOT-HEART trial found that AI-derived myocardial "radiomic phenotypes" from routine cardiac CT scans can predict long-term heart attack risk better than traditional risk factors alone. Adding these imaging signatures to standard risk models meaningfully improved predictive accuracy over an 8.6-year follow-up period.
AI reads your heart scan like never before
Researchers analyzing data from the SCOT-HEART trial found that AI-generated imaging signatures — called myocardial radiomic phenotypes — extracted from routine cardiac CT angiography (CTA) scans can predict long-term heart attack risk beyond what traditional risk scores and coronary artery assessments offer. The study followed 1,736 patients with stable chest pain for a median of 8.6 years, tracking fatal and non-fatal myocardial infarctions (MIs).
Using an open-source deep learning tool, the team identified six distinct clusters of left ventricular myocardial features — things like texture, size, and signal uniformity — and found that adding these to standard clinical models meaningfully boosted predictive power. Notably, factors like male sex and hypertension were linked to specific myocardial structural patterns, while calcium scores correlated with texture changes rather than structural ones.
By the Numbers
Why it matters: Most cardiac risk tools focus on coronary artery disease and traditional risk factors. These AI-derived myocardial signatures offer a new layer of information from scans patients are already getting — potentially helping clinicians catch high-risk individuals earlier without additional testing.
