EEG and Brain-Computer Interfaces

The paper develops a semi-automated seizure detection system using interpretable machine learning models and an open-source tool for improved efficiency and accuracy.

SiamEEGNet uses a Siamese neural network to detect drowsiness by analyzing relative EEG changes between baseline and time windows, outperforming existing methods in cross-subject scenarios.

This paper proposes an XGBoost model for automatic sleep stage classification using Dreem headband signals, achieving 81.34% accuracy.

This paper compares two machine learning models for detecting errors in human-robot interaction using EEG data, finding significant performance differences between them.

This paper introduces a real-time system for detecting selective auditory attention using EEG microstates, showing improved classification performance with 6 optimized microstate topographies.

The paper decodes semantic content from EEG data during rapid serial visual presentation by distinguishing between image statistics and semantics, finding simpler models are more robust to statistical biases.

This study develops an olfactory-based neurofeedback system using EEG alpha rhythm, showing increased alpha power in true NFB groups compared to a mock group.

This paper introduces a dimensionality reduction approach for motor imagery BCI using functional clustering, graph signal processing, and differential evolution, achieving high classification accuracy.

This paper introduces a novel feature extraction method for Motor Imagery EEG classification using a Collective Almost Synchronization model, achieving 74.03% accuracy on a benchmark dataset.

The paper validates a wearable PSG system, ZMax, comparing its sleep scoring and metrics to standard PSG, finding moderate to substantial agreement with human scorers.

The paper uses deep learning to predict electrode locations in macaque brains from ECoG signals, finding unique temporal patterns associated with different cortical regions.

This paper compares EEG signal classification methods for motor imagery in time and time-frequency domains using SVM and LDA with PCA, finding SVM performs best.

This paper re-evaluates the benefit of EEG pre-processing using the same data as a previous study, arguing that the metric used has limited applicability and pre-processing may be critical in some cases.

The paper proposes a data processing pipeline that generates graph representations of iEEG data and evaluates their effectiveness for seizure detection using graph neural networks.

OICNet improves real-time EEG source separation by integrating non-Gaussianity and orthogonality constraints, outperforming existing methods in accuracy and efficiency.

This study compares manual and automatic artifact detection in sleep EEG recordings, finding that automatic methods can effectively eliminate artifact effects without requiring visual inspection.

This paper classifies seizure dynamotypes from surface EEG using visual and automated methods, finding bifurcation prevalence aligns with intracranial data and varies by clinical factors.

This paper introduces a method combining CSP and beamformer source reconstruction for decoding M/EEG oscillatory activity, achieving performance comparable to source space decoding with lower computational cost.

This paper introduces a framework for evaluating deep learning classifiers on raw EEG data, focusing on spatial, spectral, and temporal explainability, and applies it to MDD diagnosis.

This paper introduces a multiclass full-epoch decoding approach for M/EEG data that improves accuracy over sliding window methods while enabling interpretation of spatiotemporal features.

This study shows that trial averaging improves EEG classification performance, especially when applied to both training and testing data, outperforming Mixup in some cases.

The paper examines changes in EEG features during repetitive motor tasks and their impact on BMI performance.

This paper introduces a hidden Markov model for unsupervised recalibration of intracortical brain-computer interfaces, showing improved performance over long-term use.

This paper introduces a convolutional autoencoder-based clustering method for resting-state EEG data, identifying 8 EEG states in schizophrenia and linking time outside the dominant state to negative symptom severity.

This paper evaluates a garment-based EEG system that uses smart textiles to measure EEG activity, showing comparable performance to dry EEG systems but with better comfort and lower environmental impact.

This paper validates a low-cost EEG headband for neurofeedback and introduces a method to distinguish SMR from alpha rhythms using a single recording site.

This study shows that two BCI learning phenomena can be explained by gradient-based optimization, without requiring special mechanisms.

This study reviews 916 peer-reviewed articles using consumer-grade EEG devices, highlighting their applications and future potential in neuroscience research.