Artificial Intelligence-Enabled CMR Tissue Characterization Predicts Reverse Remodeling and Clinical Outcomes in Non-Ischemic Dilated Cardiomyopathy
Authors
Affiliations (1)
Affiliations (1)
- Seoul National University Bundang Hospital
Abstract
AimsDiffuse myocardial fibrosis contributes to adverse remodeling and heart failure progression in non-ischemic dilated cardiomyopathy (NIDCM). Quantitative cardiac magnetic resonance (CMR) tissue mapping may improve risk stratification, but manual post-processing limits clinical application. This study aimed to evaluate the prognostic and functional significance of artificial intelligence (AI)-assisted automated quantification of native T1 and extracellular volume fraction (ECV) from baseline CMR in patients with NIDCM. Methods and ResultsA total of 347 consecutive patients with NIDCM who underwent baseline CMR at two university-affiliated hospitals between 2018 and 2023 were retrospectively analyzed. An AI algorithm automatically quantified whole-myocardial native T1 and ECV. The primary endpoint was a composite of cardiovascular death or hospitalization for heart failure (HHF). Left-ventricular reverse remodeling (LVRR) was assessed using serial echocardiography. Prognostic performance was assessed using Cox regression and time-dependent receiver-operating characteristic analysis. During a median follow-up of 37.9 months (IQR: 18.1-70.1), 119 patients (34.3%) experienced CV death or HHF. Elevated ECV ([≥]30%) independently predicted adverse outcomes (adjusted hazard ratio: 3.09; 95% CI: 1.87-5.12; p <0.001), whereas native T1 ([≥]1325 ms) showed weaker predictive ability. Patients with lower ECV had significantly greater LV functional improvement, while those with higher ECV demonstrated limited LVRR. Individuals with both elevated ECV and native T1 had the lowest composite LVRR rate (52.3%). ConclusionsAI-assisted automated ECV quantification predicts both LVRR and adverse clinical outcomes in NIDCM. The integration of automated CMR tissue mapping into clinical workflows may enhance precision in risk stratification and guide individualized management.