AHA Scientific Sessions 2026 Abstract · Nov 2, 2026
Myocardial Texture Features of Cardiometabolic Risk and Cardiovascular Disease in African American Adults
By Cuellar-Lobo M, Peters K, Meng Y et al.
Background: Cardiometabolic diseases are major drivers of cardiovascular morbidity, particularly among African Americans (AA). Advanced image analysis (radiomics) may detect microstructural alterations of the left ventricular myocardium (LV). Purpose: To identify myocardial radiomic CMR features associated with cardiometabolic risk factors in a community adult AA sample. Methods: We studied 1,211 participants from the Jackson Heart Study Exam 3 (2009-2012) with complete CMR and clinical data. From 5 LV short-axis cine SSFP images spanning base to apex, mean radiomic features were extracted using PyRadiomics. Features included first-order intensity statistics and five texture families: gray-level co-occurrence matrix (GLCM), gray-level run-length matrix (GLRLM), gray-level size-zone matrix (GLSZM), gray-level dependence matrix (GLDM), and neighborhood gray-tone difference matrix (NGTDM), characterizing spatial heterogeneity of myocardial signal intensity. Reproducible features with high within-family correlation (|r| ≥ 0.80) or collinearity (VIF ≥ 10) were excluded; one feature per family was selected by highest age- and sex-adjusted partial correlation with CVD, yielding 6 features. Associations with cardiometabolic risk factors [body mass index, SBP, fasting blood glucose (FBG), and total/HDL cholesterol ratio (TC/HDL)] were evaluated by linear regression adjusting for age and sex, LV mass index, end-diastolic volume, and ejection fraction, and prevalent CVD / heart failure. Benjamini-Hochberg FDR correction was applied. Results: Demographic characteristics are presented in the Table. After FDR correction, higher BMI, SBP, FPG and TC/HDL ratio, were associated with radiomic features of greater myocardial signal heterogeneity, including higher First-order, GLCM, GLRLM, GLSZM, and GLDM feature values and lower NGTDM coarseness (Figure). BMI, FBG, and TC/HDL showed the strongest associations with First-order Energy, and SBP with GLDM Gray Level Non-Uniformity. Conclusions: In AA adults in the community, cardiometabolic risk factors are associated with myocardial texture features of myocardial heterogeneity on non-contrast CMR. Our findings suggest radiomics features capture systemic cardiometabolic burden beyond conventional imaging measures.