Search Results (267)

Portable biosensor hardware can now sustain continuous multimodal physiological acquisition at the edge, yet the analytical layer that converts raw signals into deployment-consistent inference remains the main bottleneck for practical embedded systems. This study addresses that bottleneck by presenting the machine-learning layer of the Real-time Cognitive Grid, the analytical companion to the previously reported hardware architecture, which equips a fixed-wiring biosensor assembly with real-time physiological-state classification through an asymmetric edge–cloud workflow. The proposed framework assigns analytical responsibility across tiers: a locked 17-feature schema comprising 5 EMG features, 6 EEG spectral features, 2 cross-modal features, 2 HRV features, 1 EOG feature, and 1 EEG quality indicator governs window-bounded inference on the Arduino Nano RP2040 Connect with an LDA edge artefact requiring approximately 716 B RAM, whereas the cloud tier supports public-dataset pretraining, hardware-aligned refinement, multimodal fusion, deployment comparison, and feature-importance analysis under the same schema contract. To evaluate analytical consistency across physiological diversity, five public repositories covering stress physiology (WESAD), affective EEG (DEAP), inertial activity recognition (PAMAP2), sEMG gesture decoding (EMG Gestures), and motor-imagery EEG (EEGMMIDB) were evaluated under subject-disjoint GroupKFold (k = 5) protocols. To test whether the same contract survives translation to the physical rig, the hardware branch was evaluated under session-disjoint GroupKFold across five bench-acquired sessions. Unimodal performance was strongest in sEMG- and IMU-dominant tasks, whereas multimodal fusion improved macro-F1 by up to 0.141 over the strongest unimodal baseline in WESAD and by 0.109 in PAMAP2. In the hardware branch, the deployed edge LDA artefact reached 0.9435 macro-F1 with 0.9470 accuracy, while the retained cloud Random Forest reached 0.8792 macro-F1 with 0.8799 accuracy; feature-importance analysis further showed that the final 17-feature branch was dominated by EMG descriptors, with EEG spectral terms contributing secondary support and hardware-exclusive variables remaining weak under the present bench regime. These results show that a compact multimodal sensing assembly can be elevated beyond passive signal capture into an intelligent portable biosensor that performs context-aware interpretation with minimal user intervention, supported by a reproducible analytical workflow that remains coherent across heterogeneous benchmark repositories, hardware-specific refinement, and microcontroller-class deployment, thereby establishing cross-session bench feasibility as a structured basis for future multi-subject wearable validation. Full article

(1) Background: Motor imagery (MI) analysis via electroencephalography (EEG) is crucial for Brain–Machine Interfaces (BMIs) and neurorehabilitation, though inherent EEG noise challenges robust decoding algorithm development for assistive devices aiding patient recuperation. (2) Methods: This study introduces a deep learning pipeline utilizing the IFNet convolutional neural network with frequency filtering and two scaling methods to decode pedaling MI. To mitigate catastrophic forgetting during transfer learning, a rotative Elastic Weight Consolidation (rEWC)-based weight adjustment procedure was incorporated. The pipeline was validated with 23 non-disabled participants across four sessions each, evaluating four approaches via pseudo-online (open-loop) and real-time online (closed-loop) analyses. (3) Results: The decoding accuracy for MI exceeded 80% (up to 95% individually) in selected pseudo-online approaches, while real-time online classification averaged 70% accuracy with a slight bias towards the relax task, confirming statistical significance for both scenarios. (4) Conclusions: These findings demonstrate that mitigating catastrophic forgetting in transfer learning within deep learning models and applying proper scaling are crucial for optimal performance. Consequently, the proposed deep learning pipeline holds significant promise for effective, adaptive EEG-based neurorehabilitation systems. Full article

Motor imagery-based brain–computer interfaces (BCIs) have attracted significant attention due to their potential applications in assistive technologies, neurorehabilitation, and wearable human–machine interaction systems. However, practical implementation of EEG-based BCIs remains challenging because high-density EEG recordings increase hardware complexity, prolong setup time, and introduce substantial spatial redundancy. In addition, the performance of motor imagery classification strongly depends on preprocessing strategy and temporal segmentation parameters. Unlike previous studies that primarily focused on developing new deep learning architectures, this work proposes a systematic optimization framework for identifying practical low-channel EEG configurations by jointly analyzing channel selection, temporal segmentation, and preprocessing strategies for subject-independent motor imagery classification. This study investigates the influence of EEG channel reduction, temporal window segmentation, and baseline correction on motor imagery classification performance using EEGNet-based deep learning architectures. Experiments were conducted using the publicly available PhysioNet EEG Motor Movement/Imagery dataset under subject-independent evaluation conditions. Several EEG configurations were analyzed, including full-scale 64-channel recordings and reduced 15-, 6-, 3-, and 2-channel motor-cortex setups. The obtained results demonstrate that reduced-channel EEG configurations can achieve performance comparable to full-scale recordings. The best classification accuracy of 65.04% was achieved using a 15-channel motor configuration combined with 2 s sliding-window segmentation and baseline correction, achieving performance comparable to the conventional 64-channel setup (64.76%), while substantially reducing the number of electrodes and hardware complexity. Statistical analysis confirmed that the difference between the two configurations was not significant (paired t-test, p = 0.1684). Furthermore, compact 3-channel configurations maintained classification accuracy above 60%, supporting the feasibility of lightweight wearable EEG systems for practical BCI applications. The experiments additionally revealed that shorter temporal windows improve classification stability and reduce susceptibility to unrelated background EEG activity. Baseline correction significantly improved model generalization by compensating for inter-trial signal variability and slow EEG drift. Overall, the findings of this study demonstrate that careful optimization of electrode selection and preprocessing strategies can substantially improve the practicality of lightweight EEG-based motor imagery classification systems while reducing hardware complexity and preserving competitive performance. Full article

Brain–computer interface (BCI) systems have demonstrated significant potential across medical, educational, and entertainment domains. Recently, visual imagery (VI) has emerged as an alternative to traditional motor imagery (MI) paradigms, offering a broader spectrum of control signals for dexterous assistive devices. In this study, we propose a novel BCI framework for classifying visually imagined arrows defined by different colors and directions. The proposed framework employs the Choi–Williams time–frequency distribution (CW-TFD) to construct a joint time–frequency–spatial representation (TFSR) of EEG signals. The resulting TFSR is converted into grayscale images and provided as input to a newly designed convolutional neural network (CNN), which performs 16-class decoding of visually imagined arrows defined by combined color and direction attributes. A new EEG dataset was collected from 16 subjects who imagined 16 distinct arrows comprising four colors and four directions. The framework achieved an average classification accuracy of 95.05% and a Cohen’s kappa score of 0.947 across the 16 classes. To comprehensively evaluate the proposed approach, three comparative analyses were conducted. First, multiple time–frequency representations were assessed for VI-based EEG decoding. Second, the proposed CNN architecture was benchmarked against several state-of-the-art pre-trained deep learning models. Third, the framework was compared with conventional machine learning classifiers using handcrafted features. Results demonstrate that the constructed CWD-based TFSR combined with the proposed CNN consistently outperforms alternative representations and classification models. These findings demonstrate the feasibility of decoding an expanded set of visually imagined color–direction arrow commands in a subject-specific EEG-based BCI setting, supporting further development of calibrated VI-based BCI systems for assistive and interactive applications. Full article

Motor Imagery-based brain computer interface (MI-BCI) systems capable of decoding imagined movements and their kinematics are a rapidly advancing area of BCI research. Such BCIs can enhance human-computer interaction and have potential neurorehabilitation and assistive technology applications. This study explores the feasibility of decoding kinematic information, including movement direction and speed of imagined hand movements, from EEG slow cortical potentials (SCPs). EEG data from fourteen healthy subjects, associated with bidirectional center-out right-hand movement imaginations at two different speeds, is analyzed in this study. Peak negativity of movement-related cortical potential derived from fifteen primary motor cortex EEG channels is used to decode the direction and speed of imagined and observed hand movements. A Pearson correlation coefficient-based channel selection is further applied to identify a subject-specific set of channels from the pool of fifteen channels for decoding the kinematic information. Pairwise classification of direction-speed combinations achieved an average accuracy of 63.44 ± 9%. In contrast, slow-versus-fast speed classification achieved a lower accuracy of 53.87 ± 6.4% for motor imagery, which was not significantly different from the empirical chance distribution. The same analysis applied to movement observation resulted in an average direction-speed pair classification accuracy of 57.74 ± 8.6%, while speed classification achieved 50.74 ± 8.1%. These findings demonstrate that SCP features contain reliable information related to movement direction, whereas speed-related information appears weaker and less consistent across subjects. The results highlight the potential of SCP-based decoding for directional control and motivate further investigation of speed-related neural signatures. The findings from direction decoding during movement observation open avenues for future investigations into shared neural representations underlying passive movement observation. Full article

Brain–computer interface (BCI) technology has shown potential for future rehabilitation-related and assistive control applications. Nevertheless, single-modality electroencephalography-based motor imagery (EEG-MI) signals are susceptible to interference, whereas existing algorithmic models suffer from limited classification accuracy and insufficient actionable control commands for interactive devices, thereby impeding their practical deployment. To tackle these limitations, this study presents a multimodal human–computer interaction control scheme that integrates eye-movement command encoding with EEG motor imagery decoding. Self-collected EEG-MI and eye-movement datasets were established to support the proposed multimodal control framework. In this framework, eye movements are not used merely as auxiliary inputs, but are encoded as discrete commands for start, stop, grasp, and release, thereby reducing the command burden of EEG-MI decoding. The EEG-TransNet model is enhanced by integrating a time–frequency feature branch and replacing the original convolutional encoder with an adaptive multi-branch EEG feature gating module, strengthening the representation and fusion of multi-domain features. The model yields average classification accuracies of 86.96% and 88.73% on the BCI IV-2a dataset and the self-collected EEG dataset, respectively. Four independent SVM binary classifiers are adopted to identify four eye movement patterns. The EEG and eye movement classification results are binary-encoded to generate hardware-compatible control commands. Robotic-arm grasping experiments with healthy trained participants showed an average task completion time of 17 s, and the repeated grasping success-rate results further provide preliminary evidence for the real-time feasibility of the multimodal control framework under controlled laboratory conditions. Full article

Decoding mental states from non-invasive neural recordings is central to brain-computer interface research. Multimodal acquisition that combines electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) couples the high temporal resolution of EEG with the spatial specificity of fNIRS, compensating for the individual limitations of each modality. While such hybrid systems achieve strong intra-subject performance, cross-subject generalization remains constrained by inter-individual variability in neural responses. This study introduces the Eigenvalue-Guided Positional Encoding Classification Transformer (EPECT), an architecture that integrates eigenvalue-aware multi-head self-attention with sinusoidal positional encoding to capture both the spectral structure of the learned feature representations and the temporal ordering of multimodal sequences. Stacked one-dimensional convolutions extract local patterns prior to transformer encoding, and global average pooling aggregates the final representation for classification. EPECT was evaluated on two publicly available EEG-fNIRS datasets covering motor imagery (MI), n-back, discrimination/selection response (DSR), and word generation (WG) paradigms under a cross-subject protocol. The model achieved classification accuracies of 97.3%, 96.3%, 98.1%, and 97.9% on the MI, n-back, DSR, and WG tasks, respectively. Ablation studies quantified the contribution of each architectural component, and integrated gradients analysis revealed structured modality-specific attribution patterns aligned with task-relevant cortical regions. Additional experiments with synthetic cortical perturbations demonstrate the sensitivity of EPECT to subtle activity changes, indicating potential utility for tracking neurorehabilitation outcomes in future clinical applications. Full article

Interpretability is essential for deploying deep learning (DL) models in electroencephalography (EEG)-based neurotechnology, particularly in brain–computer interfaces and clinical decision-support settings. Existing post hoc explainable artificial intelligence (XAI) methods often yield single-domain attribution maps, limiting their capacity to characterize the joint spatial, temporal, and spectral structure of EEG dynamics. In addition, perturbation-based strategies may disrupt physiological signal organization, whereas gradient-based methods require access to model internals and are therefore tied to specific classifier architectures. Here, we introduce space–time–frequency KernelSHAP (STF-KernelSHAP), a model-agnostic Shapley framework for physiologically coherent EEG explainability. The method comprises three stages. First, EEG trials are decomposed into structured channel–time–frequency cells using segment-wise spectral analysis. Second, coalitions are formed over complete channel–time–frequency cells and reconstructed in the signal domain to support physiologically informed perturbations. Third, class-conditional relevance is estimated with a KernelSHAP-based weighted surrogate model that uses only model outputs, enabling architecture-independent Shapley estimation. We evaluate STF-KernelSHAP on two prerecorded public datasets: the GIGA motor imagery/movement execution (MI-ME) dataset for motor imagery (MI) decoding and the IEEE DataPort EEG Data for Attention-Deficit/Hyperactivity Disorder (ADHD)/Control Children dataset for ADHD detection. For ADHD detection, the T-GARNet base classifier interpreted with STF-KernelSHAP achieved 73.33% accuracy and 79.86% area under the curve (AUC); these values characterize classifier performance rather than the explainer itself. We compare the framework against KernelSHAP, local interpretable model-agnostic explanations (LIME), Occlusion, Integrated Gradients, and gradient-weighted class activation mapping++ (Grad-CAM++). Fidelity is assessed with Deletion and remove and debias (ROAD), while qualitative analyses examine topographic and frequency-band attribution maps. Results show that STF-KernelSHAP remains functionally competitive with established XAI methods while providing window-dependent and frequency-specific explanations. Overall, STF-KernelSHAP offers a physiologically informed and model-agnostic alternative for multidomain EEG interpretability. Full article

Decoding short-window electroencephalography (EEG) signals is critical for low-latency brain–computer interfaces (BCIs), yet current models struggle to extract robust features under high cross-subject variability and low signal-to-noise ratios. To address this, we propose a spatiotemporal decoding framework integrating dynamic weighted permutation entropy (DWPE) with a hybrid neural network. We introduce DWPE to quantify nonlinear dynamic complexity while retaining amplitude information. These features are subsequently processed by a cascaded convolutional neural network (CNN) and bidirectional long short-term memory (BiLSTM) architecture with spatial attention, enabling the simultaneous extraction of topological patterns and temporal dependencies. The framework was evaluated on three public motor imagery datasets (hBCI, BCI Competition IV-2a, and IV-2b) using a fixed 3 s window. Empirical results demonstrate that our approach achieves an average accuracy of 84.35% and an AUC of 0.8821 on the hBCI dataset, significantly outperforming current representative recent baselines (p < 0.01). Ablation studies confirm that integrating DWPE yields a 3.89% accuracy improvement over the spatial–temporal backbone alone. With a single-sample inference time of 20.94 ms and an estimated total decision latency of approximately 3.02 s under the 3 s window setting, the proposed method provides a favorable balance between decoding accuracy and computational efficiency for short-window and near-online BCI applications. Full article

Electroencephalography (EEG) is widely used in brain-computer interfaces (BCIs) for decoding motor imagery (MI) signals. However, existing methods remain limited in extracting multi-scale local spatiotemporal features and effectively integrating them with global feature information, leaving room for further improvement in classification accuracy. To address this issue, we propose HMT-Net, a hybrid architecture that integrates multi-scale convolution, the Mamba state-space model, and a self-attention mechanism. The model consists of a shallow feature embedding (SFE) module for spatiotemporal feature extraction, a multi-scale local feature extractor (MSLFE), and a Mamba–transformer global feature encoder (MTGFE). Specifically, the MSLFE employs dual-branch convolutions and channel attention to achieve adaptive multi-scale perception, while the MTGFE combines Mamba’s linear sequence modeling capability with multi-head attention to efficiently capture global dependencies. Unlike conventional Mamba or transformer EEG models, HMT-Net couples linear state-space modeling with global pairwise attention, avoiding the representational limits inherent in each individual architecture. Experiments on the BCI-IV-2a, BCI-IV-2b, and HGD datasets show that HMT-Net achieves subject-dependent accuracies of 84.07%, 89.60%, and 96.02%, respectively, outperforming EEGNet, FBCNet, EEGConformer, and ATCNet by 11.65%, 5.02%, 5.13%, and 6.60%, respectively, on BCI-IV-2a. Furthermore, HMT-Net achieves the best accuracy in subject-independent experiments, demonstrating strong generalization capability. Ablation studies and visualizations further validate the effectiveness and interpretability of the proposed model. Full article

Optically Pumped Magnetometers (OPMs) have emerged as a promising technology for developing flexible, wearable magnetoencephalography (OPM-MEG) systems, offering high spatiotemporal resolution without the need for cryogenic cooling. However, their application to phase–amplitude coupling (PAC)-based neural decoding remains largely unexplored. Investigating their decoding performance is essential for evaluating the capability of OPM-MEG in characterizing complex neural dynamics and discriminating motor states. In this study, OPM-MEG was utilized to record brain activity during rest, motor imagery, and motor execution tasks. A two-stage temporal optimization strategy combining time-resolved PAC localization and the Kullback–Leibler modulation index (KL-MI) was employed to extract robust PAC features from low-frequency phase and high-frequency amplitude coupling. α–γ and θ–γ PAC features were subsequently fed into a multiclass linear discriminant analysis (LDA) classifier for motor state decoding, and compared against baseline band-power feature decoding performance. Experimental results demonstrate that PAC features derived from OPM-MEG significantly outperform the corresponding baseline band-power features in decoding performance. Notably, α–γ PAC features effectively discriminate among different motor states, achieving a balanced accuracy of 85.91% in 10-fold cross-validation. This performance significantly exceeds the 50% one-vs-rest chance level and outperforms θ–γ PAC features. These findings provide initial evidence for the feasibility of OPM-MEG in PAC-based motor state decoding and a preliminary case study for characterizing motor-related neural dynamics in a wearable MEG system. Full article

The practical application of Brain–Computer Interface (BCI) technology is frequently challenged by significant inter-individual variability in electroencephalogram (EEG) signals. This variability makes it extremely difficult to decode the brain activity of new subjects using pre-recorded data from previous subjects. To address these issues, this study presents an EEG decoding approach based on four-stage domain generalization. We start by preprocessing the data and then dividing it into source and target domains. The source domain data are then passed through four sequential modules: Feature Extraction, Feature Augmentation, Feature Optimization, and Domain Adaptation, where we adjust the parameters using the source domain loss function. Next, the target domain data go through the same four stages while we fine-tune the parameters together with the domain adaptation loss, ultimately obtaining the decoding results for the target domain. The proposed method achieves the highest classification accuracy of 72.61%, outperforming EEGTransferNet by 7.22% and surpassing all classical and deep learning baselines by improvements ranging from 5.97% to 23.86%. Overall, the proposed method significantly enhances cross-subject generalization in motor imagery decoding, offering practical value for plug-and-play BCI applications. Full article

Motor imagery brain–computer interfaces (MI-BCIs) have important applications in neurorehabilitation, assistive communication, and non-muscular human–machine interaction. From a bionic neural-interfacing perspective, MI-BCI decoding provides a computational bridge between biological motor intention and external machine control. However, reliable motor imagery electroencephalography (MI-EEG) classification remains challenging due to the highly non-stationary features of MI-EEG and limited interpretability. In this work, we propose PG-MCTFormer, a prior-guided multi-scale convolutional Transformer for MI-EEG classification that integrates rhythm-aware temporal filtering, dual-scale spatial modeling, and contextual decoding within a unified architecture. We evaluated the model on the publicly available BCI Competition IV 2a dataset, achieving 85.08% average accuracy and a Cohen’s kappa of 0.80, with significant performance improvement over the traditional methods. Comprehensive multi-view interpretability analyses in the frequency, temporal, and spatial domains further show that the learned filters remain aligned with canonical MI-related bands, discriminative evidence concentrates in the middle-to-late imagery interval, and the spatial prior is refined into subject-adaptive sensorimotor topographic patterns. These results indicate that explicit neurophysiological priors can improve both the robustness and the interpretability of MI-EEG decoders for biomimetic neural-interface applications. Full article

Existing deep-learning-based motor imagery (MI) electroencephalogram (EEG) decoding methods face challenges in generalizing across sessions and providing channel-level physiological interpretability. These limitations hinder the practical application of MI-EEG systems. Accordingly, DO-PI-EATCNet (Dream-Optimization-Enhanced, Physics-Inspired, Efficient-Attention Temporal Channel Network) is proposed to improve generalization and interpretability in MI-EEG classification. Unlike models that simply combine multiple components, DO-PI-EATCNet assigns distinct roles to feature representation, temporal channel modeling, temporal regularization, and channel compactness. Latent-Projected Attention (LPA) enhances spatiotemporal discriminability by aligning attention in a low-dimensional latent space, and Temporal Channel Cascaded Collaborative Attention (TCCA) refines dependencies between time and channels. Fractional-Order Difference Temporal Consistency Loss (FD-TCL) is introduced as a neurodynamics-inspired temporal regularizer to reduce high-frequency fluctuations in prediction sequences and improve within-subject cross-session prediction stability. The Multi-Population Dream Optimization Algorithm (MPDOA) is used for channel selection to obtain a compact EEG channel subset and reduce computational load, although it introduces a slight accuracy decrease compared with the uncompressed full model. Under a within-subject cross-session protocol on the BCI Competition IV-2a four-class MI dataset, the final compact model achieves an average accuracy of 84.4% and Cohen’s κ of 0.790, outperforming the reimplemented baselines. Compared with the uncompressed LPA-TCCA-FD-TCL variant, MPDOA slightly decreases accuracy from 84.9% to 84.4%, but reduces EEG channels from 22 to about 15 and decreases MACs by 27%. Scalp topographies and selected-channel visualizations provide qualitative support for channel-level anatomical plausibility, as the selected electrodes are mainly located over expected sensorimotor-related regions, while t-SNE offers a descriptive visualization of the learned feature distributions. Full article

Motor imagery (MI)-based brain–computer interfaces (BCIs) have led to great interest as a result of their potential use in neurorehabilitation, assistive robotics, and human–computer interaction. However, decoding electroencephalographic (EEG) signals with high accuracy continues to be a difficult task due to the weak signal-to-noise ratio, differences among subjects, and the complicated temporal–spectral–spatial neural dynamics. Deep learning methods recently developed, such as convolutional neural networks, recurrent architectures, graph neural networks, and adversarial transfer learning, have enhanced MI decoding performance, yet many models are still concentrating on a single representation domain or they need costly adaptation phases in terms of computation. To tackle these shortcomings, we present NeuroCrossNet, a unified tri-modal deep learning model that is able to learn the temporal, spectral, and spatial EEG features jointly for robust and calibration-free MI decoding. The suggested network combines a Temporal HyperMixer Block for capturing long-range temporal dependencies, a wavelet transformer for learning localized time–frequency representation, and a Graph Attention Network for EEG topology-aware spatial reasoning. Additionally, a Dynamic Residual Attention Gate (DRAG) has been developed to adaptively merge heterogeneous feature streams, and a compact subject-aware normalization (SAN) method enhances cross-subject generalization without the use of labeled target-domain calibration data. Our proposed model was tested following the rigorous leave-one-subject-out (LOSO) approach on BCI Competition IV-2a and High-Gamma datasets. NeuroCrossNet reached a classification accuracy of 91.30%, surpassing several strong benchmark methods, including CNN-LSTM, EEGNet, DeepConvNet, spectral CNN, and graph-based EEG decoding frameworks. Furthermore, a large number of ablation studies reveal that the integration of temporally, spectrally, and spatially complementary representations considerably boosts robustness and inter-subject consistency. Full article