Waveband-Resolved Nonlinear Dynamics and Cross-Phase Coupling of SERF-Magnetocardiography Signals via Phase-Space Reconstruction

Abstract

Background: Spin-exchange relaxation-free magnetocardiography (SERF–MCG) offers femtotesla-level sensitivity for noninvasive cardiac monitoring. However, conventional analysis pipelines, largely adapted from linear electrocardiography, often fail to capture the subtle nonlinear dynamical perturbations induced by myocardial ischemia. Ischemia is hypothesized to disrupt the intrinsic coordination between depolarization and repolarization, necessitating a phase-resolved nonlinear characterization. Methods: We developed a waveband-resolved framework that decomposes each SERF–MCG beat into P, QRS, ST, and T segments, reconstructing their segment-specific attractors via delay-coordinate embedding. From these topological structures, we extracted a 30-dimensional feature set capturing geometric, entropic, chaotic, and recurrence properties. Intra-beat interactions were assessed using a dual-perspective approach: (i) statistical coupling matrices derived from feature aggregates, and (ii) an interpretable four-node self-attention encoder that estimates a 4 × 4 matrix of model-derived associations between cardiac phases. Results: Analysis of 1,848 statistically independent median beats (906 controls, 942 clinically diagnosed ischemia patients) showed statistically significant but partially overlapping nonlinear feature differences between groups. Specifically, repolarization-related ST/T features were associated with lower sample entropy (p < 10−8) and higher Lyapunov-related descriptors in ischemic recordings, consistent with reduced complexity and increased local instability. Both statistical and attention-based analyses showed increased ST–T association in ischemic recordings, suggesting altered dependence between early and late repolarization phases. Conclusions: This study suggests that myocardial ischemia in SERF–MCG is associated with reduced repolarization complexity, increased local instability, and stronger ST–T association, rather than purely morphological deviations. By combining phase-space reconstruction with interpretable attention mechanisms, the proposed framework provides complementary, physiologically interpretable signal markers that may support future noninvasive ischemia assessment after prospective clinical validation.