AHA Scientific Sessions 2026 Abstract · Nov 2, 2026
Association of Myocardial Texture with Cardiometabolic Risk and Heart Failure
By Zimmermann S, Dong Z, Cuellar-Lobo M et al.
Background: Heart failure (HF) often develops after prolonged exposure to cardiometabolic risk factors. Cardiac magnetic resonance (CMR) radiomics may provide insight into myocardial tissue texture patterns that are not captured by conventional measures of left ventricular (LV) structure and function. Research Question: We aimed to determine whether cardiometabolic risk factors and HF are associated with radiomic myocardial tissue characteristics in adults in the community. Methods: In 1,008 Framingham Heart Study 2 nd Generation participants who underwent non-contrast 1.5T CMR (Philips Achieva), LV myocardial texture analysis by radiomics were extracted from end-diastolic short-axis cine images (PyRadiomics, Python) and normalized for comparison across radiomics families. Using published reproducible cine CMR features, we eliminated those with high correlation (r≥0.8) and collinearity (VIF≥10) and selected eight top-ranking features from each feature family correlated with coronary heart disease. We examined the association of this parsimonious group of radiomics features with contemporaneously measured (Exam 8) markers of cardiometabolic risk including body mass index (BMI), systolic blood pressure (SBP), fasting glucose, total cholesterol-HDL ratio, and prevalent HF. We conducted multivariable regression models adjusted for age and sex, medication treatment for hypertension, diabetes, and hyperlipidemia, LV mass index, end diastolic volume index, and ejection fraction, and prevalent cardiovascular disease. Results: Adverse cardiometabolic risk factors, including higher BMI, higher total cholesterol-to-HDL ratio, elevated fasting glucose, and higher SBP, were associated with radiomic features reflecting greater myocardial texture heterogeneity and complexity, including higher first-order energy and gray-level co-occurrence matrix features, and lower neighborhood gray-tone difference matrix features ( Table/Figure ). Obesity, elevated SBP, dyslipidemia, and HF were also associated with higher gray-level size-zone matrix non-uniformity, suggesting greater heterogeneity in zones of similar myocardial signal intensity. Higher SBP showed additional associations with gray-level dependence and run-length matrix features, reflecting greater variability in adjacent-pixel texture and linear gray-level patterns. Conclusion: Non-contrast CMR radiomics is able to identify unfavorable patterns of myocardial architecture, independently of clinical measures of LV geometry and function, associated with key markers of cardiometabolic risk and heart failure in a general adult community population. These findings may provide complementary insight into myocardial tissue remodeling related to adverse cardiovascular risk.