AI Transforms Standard Chest CTs into Powerful Predictors of Heart Failure and AFib

In a paradigm-shifting breakthrough for opportunistic screening, AI-driven cardiac chamber volumetry from non-ECG-gated chest CT scans—the overwhelming majority of scans performed annually—can reliably predict incident Heart Failure (HF) and Atrial Fibrillation (AF), matching the performance of specialized ECG-gated cardiac CTs, according to new research published in the American Journal of Preventive Cardiology.

The Lead: Bridging the Opportunistic Gaps in CVD Prevention

The clinical landscape is saturated with imaging; approximately 20 million non-gated chest computed tomography (CT) examinations are performed annually in the United States for indications ranging from lung cancer screening to trauma. Yet, less than 5% of all chest CTs are ECG-gated, which has historically restricted comprehensive cardiac analysis. Leveraging data from the multi-ethnic Multi-Ethnic Study of Atherosclerosis (MESA), this study establishes that AI-enabled opportunistic volumetry on these ubiquitous non-gated scans can transform them into powerful tools for cardiovascular risk assessment, potentially expanding the pool of usable imaging data for screening by nearly 20-fold without additional cost or radiation to the patient.

The Breakthrough: AI-CVD and AutoChamber

The researchers utilized AutoChamber Version 2.0 (HeartLung.AI), an FDA-approved, AI-based automated volumetry software. This AI tool is designed to overcome the core challenge of non-contrast, non-gated CT scans: poor differentiation between the blood pool and myocardium due to the absence of contrast enhancement. By rapidly and reliably segmenting cardiac chambers, the AI-CVD platform extracts subclinical markers, enabling a sophisticated phenotypic assessment that was previously manual and labor-intensive. The study design elegantly compared these AI-derived non-gated measurements against the gold standard ECG-gated scans acquired on the same day from the same individuals, and further validated its reliability against paired ECG-gated scans.

Key Findings: Statistical Parallels and Outcome Prediction

The study cohort included 2,053 MESA Exam 5 participants (mean age ~70 years), followed for HF over a median of 8.1 years and for AF over a median of 7.1 years.

Diagnostic Agreement on Non-Gated vs. ECG-Gated

The agreement between the two scan types for detecting chamber enlargement was exceptional, rivalling the consistency observed between paired gated scans.

Note: ICCs for all continuous volumetric metrics were $\ge0.89$, indicating excellent to good reproducibility across acquisition types.

Comparative Predictive Performance (C-indices)

Harrell’s C-indices, adjusted for age and sex, confirmed that non-gated parameters are non-inferior to ECG-gated metrics in discriminating incident HF and AF outcomes.

Furthermore, cumulative incidence curves stratified by quartiles for both HF and AF outcomes showed substantial overlap between non-gated and gated scans, visually confirming the statistical concordance. Hazard ratios per one standard deviation increase were also closely concordant for all outcomes, including all-cause mortality.

Clinical Implications: A Scalable Pathway for Preventive Cardiology

«These findings highlight the potential of AI-based cardiac volumetry across heterogeneous imaging conditions and establish a foundation for opportunistic cardiac risk assessment using the large and growing repository of existing non-gated chest CT scans,» the authors concluded.

For busy clinicians, this translates into immediate actionable data: routine lung or trauma scans can now flag patients—often asymptomatic or with underutilized risk factor assessments—who are at high risk for HF or AFib. By validating the reliability of non-gated scans, this research removes common barriers such as the need for dedicated operators, specialized imaging protocols, and high out-of-pocket costs, paving the way for a more equitable, widespread, and cost-effective screening landscape in preventive cardiology.

Citation

Mirjalili SR, Atlas K, Reeves AP, Zhang C, Wasserthal J, Azimi A, Hashemi A, Mozafarybazargany M, Jolfayi AG, Atlas T, Henschke CI, Yankelevitz DF, Zulueta JJ, Fan W, Mechanick JI, Branch AD, Nasir K, Fayad Z, McConnell MV, Rana JS, Vliegenthart R, Maron DJ, Narula J, Budoff MJ, Mehran R, Williams Kim A Sr, Shah PK, Mechanic O, Agatston AS, Kloner RA, Wong ND, Naghavi M. Heart Failure and Atrial Fibrillation Prediction from Non-gated Chest CT Scans Using AI-Based Cardiac Chamber Volumetry: An AI-CVD Study within the Multi-Ethnic Study of Atherosclerosis (mesa). American Journal of Preventive Cardiology. 2026. doi: https://doi.org/10.1016/j.ajpc.2026.101732