Cardiac CT and Machine Learning Predict Functional Improvement After AF Ablation
Published in Artificial Intelligence, Journals, Scanners and Imaging, Tools.
A multicentre study published in BMC Medical Imaging on September 28, 2026 explored whether cardiac CT angiography and machine learning can predict functional improvement after atrial fibrillation ablation in patients who also have heart failure.
Researchers combined measurements of left atrial and pulmonary vein shape, imaging features extracted from the left atrial wall, and clinical factors. The study included 240 patients for training, 101 for validation, and 75 for testing, with the test data coming from two other institutions.
The combined model achieved areas under the receiver operating characteristic curve of 0.866, 0.803, and 0.845 in those respective groups. These values measure discrimination between outcomes; they are not percentages of patients correctly diagnosed. SHAP analysis helped identify which imaging features contributed most to the predictions.
The retrospective findings suggest a possible decision-support approach, but do not establish that using the model improves treatment selection or patient outcomes. The authors say prospective evaluation is needed before clinical implementation.
Source: Jing and colleagues, BMC Medical Imaging (September 28, 2026).