Opportunistic volume assessment of left ventricular volume from coronary CT angiography: agreement with cardiac MRI

Introduction Heart failure frequently remains undiagnosed until advanced stages. Automated extraction of cardiac chamber volumes from routine CCTA may enable opportunistic identification of patients at risk of ventricular remodeling without additional imaging or radiation. However, large-scale validation studies against cardiac magnetic resonance imaging (CMR) remains limited. Aim To evaluate the agreement betweegn automated left ventricular (LV) volume measurements derived from routine coronary CT angiography and CMR in a large clinical cohort. Methods In a retrospective cohort, including patients from the Dan-NICAD 1 and 2 study, who underwent both CCTA and CMR were included, yielding 508 paired examinations LV cavity volumes were automatically quantified from CCTA using an AI-based segmentation framework, without manual editing of contours. CMR-derived LV end-diastolic volume, based on local site read, served as the reference standard. Agreement was assessed using Bland–Altman analysis, linear regression, and evaluation of proportional bias. Results In total, 508 patients (337 (66.3%) male and 171 (33.7%) females, mean age 63.1) were included. Mean LV volumes derived from CCTA and MRI were comparable across the cohort. Bland–Altman analysis demonstrated a mean bias (−0.8 mL) with limits of agreement of −52.2 to +50.7 mL. Linear regression showed a strong association between modalities (R² = 0.60, r = 0.77, p < 0.001). A significant proportional bias was observed, with increasing underestimation of LV volume by CCTA at higher volumes (slope −0.147, p < 0.001). Conclusion Automated LV volume estimation from routine CCTA demonstrates moderate to good agreement with cardiac MRI across a large clinical population, with minimal overall bias. Differences in cardiac phase definition between CCTA and MRI, with CCTA reconstructed within a late-diastolic window rather than a fixed end-diastolic time point, may contribute to the observed proportional bias. These findings support the potential role of opportunistic CCTA-based LV volume assessment as a scalable imaging biomarker for cardiac remodeling and heart failure risk stratification.