A novel interpretable and real-time framework for multichannel MCG signal quality assessment

Multichannel magnetocardiography (MCG) quality assessment is critical for ensuring diagnostic reliability, suppressing misleading noise patterns, and improving the performance of downstream analytical tasks. However, clinical MCG recordings are frequently degraded by complex environmental interference, making real-time and interpretable signal quality evaluation a significant challenge. This study proposes a lightweight and interpretable framework named Rhythm and Spatial-Enhanced Quality (RSEQ) for real-time evaluation of the quality of multichannel MCG signal. The RSEQ framework employs a two-stage rule-based strategy: rhythm consistency screening via Kullback–Leibler divergence and spatially informed signal noise ratio estimation using independent component analysis and wavelet-Wiener filtering. On the semi-simulated dataset, RSEQ achieves an accuracy of 94.07%, F1 score of 0.94, recall of 0.94, precision of 0.95, and a processing speed of 7.18 × 10³ PPS, outperforming several baseline methods. Robustness is further validated under CT-induced noise, where the method maintains a classification accuracy of 95.98%. To assess its clinical utility, RSEQ is applied to a downstream myocardial ischemia classification task. Results show that excluding low-quality segments significantly enhances diagnostic performance, with the F1 score increasing from 0.889 to 0.945 and AUC rising from 0.76 to 0.93. This confirms that quality assessment not only removes misleading signal segments but also improves the overall effectiveness of disease prediction. These findings demonstrate that the proposed framework enhances the reliability, interpretability, and diagnostic value of MCG recordings under real-world clinical conditions.