Integration of Magnetocardiography and Coronary Computed Tomography Angiography With Machine Learning for Detection of Functionally Significant Myocardial Ischemia.
Authors
Affiliations (4)
Affiliations (4)
- The Fourth School of Clinical Medicine, Zhejiang Chinese Medical University, Hangzhou First People's Hospital, 310000 Hangzhou, Zhejiang, China.
- Department of Radiology, The Second Affiliated Hospital, Zhejiang University School of Medicine, Zhejiang Chinese Medical University, 310000 Hangzhou, Zhejiang, China.
- Department of Radiology, Affiliated Hangzhou First People's Hospital, Westlake University School of Medicine, 310000 Hangzhou, Zhejiang, China.
- Centre for Intelligent Healthcare, Coventry University, CV1 5RW Coventry, UK.
Abstract
Functional assessment of myocardial ischemia is essential and can be evaluated noninvasively using coronary computed tomography angiography (CCTA) and magnetocardiography (MCG). However, the diagnostic value of integrating CCTA and MCG has not been investigated. This retrospective, single-center cohort study included 275 patients with suspected coronary artery disease (CAD) who underwent both CCTA and MCG examinations from December 2023 to June 2025. Functionally significant ischemia was defined by invasive fractional flow reserve (FFR) or CT-derived FFR (CT-FFR). Quantitative features from both modalities were extracted and normalized. Machine learning (ML) models based on MCG alone, CCTA alone, and combined MCG-CCTA were constructed and evaluated using five-fold cross-validation. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity; model interpretability was examined using Shapley additive explanations (SHAP). Of the 275 patients, 98 (35.6%) were classified as being ischemic. The MCG model achieved an AUC of 0.769 (95% confidence interval (CI): 0.708-0.829), and the CCTA model yielded an AUC of 0.755 (95% CI: 0.692-0.818). In contrast, the combined MCG-CCTA model developed using ML demonstrated superior performance, with an AUC of 0.829 (95% CI: 0.773-0.885), an accuracy of 0.800, a sensitivity of 0.704, and a specificity of 0.853. A combined MCG-CCTA model developed with ML outperforms models based on either modality alone for detecting functionally significant myocardial ischemia. In clinical practice, this integrated approach may enhance ischemia assessment and inform downstream testing decisions.