Dong Xu

Background: Spin-exchange relaxation-free magnetocardiography (SERF–MCG) offers femtotesla-level sensitivity for noninvasive cardiac monitoring. Conventional analysis pipelines, largely adapted from linear electrocardiography, may not fully capture nonlinear dynamical patterns embedded in cardiac magnetic-field signals. In this retrospective and de-identified cohort study, we aimed to develop and evaluate a waveband-resolved nonlinear analysis framework for exploratory signal-level characterization of SERF–MCG recordings from stringently screened healthy adults and a heterogeneous clinical cardiovascular patient cohort enriched for clinically diagnosed ischemia. Methods: We developed a waveband-resolved framework that decomposes each SERF–MCG beat into P, QRS, ST, and T segments and reconstructs segment-specific attractors via delay-coordinate embedding. From these reconstructed trajectories, we extracted a 32-dimensional feature set capturing phase-space geometric descriptors, entropy-related descriptors, Lyapunov-related descriptors, and spatial energy/eigenvalue-ratio descriptors. Cross-phase signal associations were assessed using a dual-perspective approach: (i) statistical association matrices derived from feature aggregates and (ii) a four-node self-attention encoder that estimates a 4 × 4 matrix of model-derived association weights between cardiac phases. Results: Analysis of 1,848 statistically independent median beats, including 906 stringently screened healthy controls and 942 recordings from a heterogeneous clinical cardiovascular patient cohort enriched for clinically diagnosed ischemia, showed statistically significant but partially overlapping nonlinear feature differences between groups. Repolarization-related ST/T features showed lower sample entropy and altered Lyapunov-related descriptors in the clinical patient cohort, consistent with group-level differences in signal complexity and local trajectory instability. Statistical and attention-based analyses further suggested altered ST–T association patterns at the signal-feature level. These findings should be interpreted as exploratory cohort-level signal differences rather than ischemia-specific causal or diagnostic evidence. Conclusions: This study presents a waveband-resolved nonlinear SERF–MCG analysis framework for exploratory characterization of cardiac magnetic-field signals. Within the present retrospective and de-identified dataset, the framework identified group-level differences in repolarization-related nonlinear features and cross-phase association patterns between stringently screened healthy controls and a heterogeneous clinical patient cohort. The findings support the methodological feasibility of phase-space reconstruction and attention-assisted association analysis for SERF–MCG signal characterization, while prospective studies with complete clinical covariates and uniform diagnostic verification are required before ischemia-specific or diagnostic claims can be made.

Non-invasive methods used for PH detection in clinical practice have several limitations. The combination of high spatiotemporal sensitivity magnetocardiography (MCG) and artificial intelligence algorithms may offer an accurate approach for PH detection. In this study, we develop a convolutional neural network (CNN) model based on the 64-channel MCG time-series data. This exploratory study enrolled patients undergoing 64-channel MCG, including right-heart-catheterization confirmed PH patients and symptomatic controls with low echocardiographic probability of PH. After data preprocessing, a temporal CNN integrating MCG signals with age, sex, and body mass index was developed and compared with conventional machine learning models. The CNN model achieved strong discrimination, with area under the curve (AUC) values of 0.939 (95% confidence interval [CI]: 0.913–0.961) in the development out-of-fold evaluation and 0.974 (95% CI: 0.944–0.994) in the hold-out test set, outperforming conventional machine learning models. Decision curve analysis showed the greatest net benefit at clinically relevant thresholds. Attribution analysis indicated that spatial QRS morphology redistribution contributed substantially to PH classification. The temporal CNN model based on raw 64-channel MCG signals showed promising performance for non-invasive PH detection and outperformed conventional machine learning approaches in this exploratory single-center cohort enriched for PAH and CTEPH.

Background: Myocardial ischemia (MI) is a major cause of morbidity and mortality worldwide and requires timely and reliable detection. Although Spin-Exchange Relaxation-Free (SERF) magnetocardiography (MCG) provides femtotesla-level sensitivity for identifying non-linear cardiac repolarization anomalies, its clinical deployment is currently impeded by the computational bottlenecks inherent to portable edge platforms. Methods: We propose a “Sensor-to-Image” Edge-AI framework that links quantum sensing with computer vision. Single-channel SERF-MCG signals from a large cohort of 2118 subjects (1135 Healthy, 983 Ischemia) were transformed into phase-space images using three distinct encoding modalities: Recurrence Plots (RP), Gramian Angular Summation Fields (GASF), and Markov Transition Fields (MTF). These visual representations were subsequently analyzed by a streamlined MobileNetV3-Small architecture, optimized for low-latency inference. To maximize diagnostic precision, an adaptive weighted fusion mechanism was engineered to combine the chaotic specificity captured by RP with the morphological sensitivity of GASF through a validation-optimized fixed global weighting strategy. Results: In our experiments, the fusion model achieved an Area Under the Curve (AUC) of 0.865, which was higher than the 1D-CNN baseline (AUC 0.857) and the single-modality models. Notably, the fusion strategy significantly elevated sensitivity to 88.3% while maintaining a specificity of 66.5%. Although specificity is moderate, this trade-off prioritizes high sensitivity to minimize false negatives in pre-hospital screening scenarios. The average inference time was 4.7 ms per sample on a standard CPU, suggesting suitability for real-time Point-of-Care (PoC) scenarios under further on-device validation. Conclusions: The results suggest that multi-view phase-space fusion can capture subtle spatio-temporal changes associated with ischemia. The proposed lightweight framework may support the development of portable SERF-MCG systems with embedded AI screening.

The heart exhibits inherently nonlinear and chaotic electrical dynamics, making the early detection of myocardial ischemia (MI) challenging using traditional electrocardiography (ECG) or standard magnetocardiography (MCG). In this study, we propose an engineering-oriented framework that integrates classical nonlinear dynamics with machine-learning-based analysis, termed the Magnetocardiography Chaotic Dynamics Map (MCDM), to reconstruct nonlinear phase-space trajectories from 36-channel MCG recordings and capture differences in reconstructed nonlinear dynamics associated with ischemic conditions. Morphological and quantitative analyses of the MCDM patterns reveal marked differences between healthy and ischemic subjects. Using a machine-learning classifier trained on HOG and LBP descriptors, the proposed MCDM-based model achieved an accuracy of 92.19%, a sensitivity of 88.75%, a specificity of 95.63%, an F1-score of 91.91%, and an AUC of 89.80%, demonstrating effective discriminative capability for early ischemia screening. Owing to its computational simplicity and noninvasive nature, the proposed MCDM framework represents a promising tool for scalable screening of ischemic heart disease.

This paper proposes Temporal-decomposition and Attention-based Denoising Network (TADNet) for magnetocardiography (MCG) signal denoising. The raw time-series data are decomposed into trend and seasonal components. The trend component is modeled using a multi-layer perceptron to capture the trend of the data. For seasonal component, a multi-layer convolutional neural network is employed to perform segmentation and extract local temporal features. These fine-grained features are processed by a Transformer-based self-attention mechanism to capture long-range temporal dependencies. Finally, deconvolution operations are applied to reconstruct the signal, achieving effective noise reduction for MCG data. This paper conducts in-depth studies on effects of time, wavelet, and Fourier style attention on model performance, enhancing the interpretability of the model. Extensive experiments on clinical MCG data under various noise conditions demonstrate that TADNet outperforms traditional algorithms, proving its effectiveness and robustness. The research results indicate that TADNet provides a highly efficient and accurate solution for MCG signal denoising.

Magnetocardiography (MCG) plays a growing role in non-invasive cardiac disease diagnosis. However, MCG signals are prone to environmental magnetic fields and metal artifacts, distorting waveforms and affecting diagnostic accuracy. Existing methods like the Fast Independent Component Analysis (FastICA) and Information Maximization (Infomax) algorithm have limitations in suppressing ultra-low frequency metal artifacts. We propose a Second-Order Blind Identification algorithm based on an optimized time-delay matrix, utilizing temporal coherence to effectively separate ultra-low frequency metal artifacts from mixed sources. An automatic screening method for metal artifacts, QRS, T/P waves, and unknown interferences is established using time-frequency features. Extensive simulations and real OPM-MCG experiments validate our method’s superiority in metal artifact suppression. The results show our method surpasses FastICA and Infomax in suppressing metal artifacts, achieving average SNR improvement of 5.36% to 29.40% across four subjects. Reconstructed P/QRS/T waves are undistorted, with a minimum 80.71% reduction in RMSE. This method potentially expands MCG’s clinical applications, benefiting more patients.

Piezoelectric beams are widely used in micro-electromechanical systems. At the microscale, the influence of the size effect on a piezoelectric beam cannot be ignored. In this paper, higher-order elasticity theories are considered to predict the behaviors of piezoelectric micro-structures and a size-dependent dynamic model of a laminated piezoelectric microbeam is established. The governing equations for the laminated piezoelectric microbeam are derived using the variational principle. The natural frequencies of piezoelectric microbeams are obtained by size-dependent dynamic models. The results reveal that the size effect can enhance the structural stiffness at the microscale. The natural frequency obtained by using the classical model is smaller than that obtained using the size-dependent model. Compared with the modified couple stress model, the modified couple stress model underestimates the size-dependent response. Thus, the modified couple stress model is a simplification of the modified strain gradient model. The influence of beam thickness on the natural frequency is also discussed. With increasing the thickness, the natural frequency of the size-dependent models gradually approaches the result of the classical model. If the value of h/l is greater than 15, the influence of the size effect can be neglected. Additionally, the relative thickness can influence the natural frequency, and if the relative thickness is greater than 5 or less than −5, the bilayer beam can be simplified to a single-layer beam.

This article studies experimentally the damage behaviors of a 59.5-m-long composite wind turbine blade under accelerated fatigue loads using acoustic emission technique. First, the spectral analysis using the fast Fourier transform is used to study the components of acoustic emission signals. Then, three important objectives including the attenuation behaviors of acoustic emission waves, the arrangement of sensors as well as the detection and positioning of defect sources in the composite blade by developing the time-difference method among different acoustic emission sensors are successfully reached. Furthermore, the clustering analysis using the bisecting K-means method is performed to identify different damage modes for acoustic emission signal sources. This work provides a theoretical and technique support for safety precaution and maintaining of in-service blades.