Opportunistic Screening for Atrial Fibrillation on Routine Chest Computed Tomography

Area of Science:

• Radiology and Cardiology
• Medical Imaging Biomarkers
• Atrial Fibrillation Detection

Background:

• Atrial fibrillation (AFib) significantly elevates stroke risk.
• Quantitative imaging biomarkers from chest CT aid in disease detection.
• Left atrial enlargement (LAE) is a known consequence of AFib.

Purpose of the Study:

• To investigate the association between AFib status and LAE on non-gated chest CT.
• To determine if CT-derived LA measurements can predict AFib.
• To evaluate the diagnostic performance of LA measurements for AFib detection.

Main Methods:

• 500 patients undergoing non-gated chest CT were analyzed.
• Left atrium maximal axial cross-sectional area (LA-MACSA) and left atrium anterior-posterior dimension (LA-AP) were measured.
• Statistical analyses and machine learning models were employed to predict AFib using LA measurements and clinical data.

Main Results:

• Patients with AFib exhibited significantly larger LA-MACSA and LA-AP compared to those without AFib (P <0.000001).
• LA-MACSA and LA-AP showed an AUC of 0.73 for AFib prediction.
• Multivariable analysis and machine learning models improved AFib prediction accuracy, with AUCs up to 0.77.

Conclusions:

• LA-MACSA (>30 cm²) and LA-AP (>4.5 cm) are specific CT-based indicators for predicting AFib risk.
• These measurements can be rapidly obtained from routine chest CT scans.
• Utilizing CT-derived LA metrics can aid in identifying patients at increased risk for AFib.