EEG and Brain-Computer Interfaces

This study shows that wetting hair with tap water improves dry electrode EEG signal quality by reducing impedance and increasing spectral and functional connectivity similarity to gel-based systems.

The paper introduces DYNAM-O, a toolbox for analyzing sleep EEG to identify individualized transient oscillations, finding sex differences in oscillatory activity.

This paper validates EEG beamforming for separating heartbeat-evoked potentials (HEP) from cardiac artefact (CA) using simulated and empirical data, showing reliable recovery and group differences in interoceptive regions.

This paper classifies expert-confirmed interictal epileptiform discharge epochs using EEG and ECG data, finding that ECG features improve categorization accuracy.

This study uses quantum machine learning models to detect sleep deprivation from EEG signals, achieving high accuracy in both epoch-level and subject-level evaluations.

The paper introduces a polar decomposition method to analyze time-odd EEG dynamics across sessions, improving cross-session subject discrimination.

Img2EEG generates synthetic EEG responses to visual inputs, generalizing across stimuli and participants while preserving neural structure and enabling targeted interventions.

The paper evaluates timing error tolerance in cognitive and neural measures across devices and introduces MERIT, a no-code platform for consistent cognitive assessment in lab and real-world settings.

This paper develops an interpretable machine learning classifier for Alzheimer's disease using resting-state EEG data from OpenNeuro, achieving high accuracy with EEG Logistic Regression.

This study investigates how grasp and object properties are encoded in the cortical grasping network of tetraplegic humans, showing separable neural representations in different brain regions that could improve brain-machine interface performance.

This paper compares Pearson correlation and mutual information estimators for analyzing neural evoked responses, identifying parameter settings and trade-offs for reliable similarity measures.

The paper uses a portable three-channel EEG system to decode futsal free-kick outcomes from pre-movement brain activity, finding session-specific neural signatures linked to performance.

This paper introduces a two-phase deep learning framework for EEG-MEG decoding in BCI, showing improved classification accuracy and transferability across participants.

This paper introduces a bilinear state-space model to track cortical traveling waves in MEG/EEG data, showing improved accuracy over traditional methods in simulations and empirical recordings.

CORSICA suppresses cochlear implant artifacts in speech-evoked EEG data using cross-correlation with speech stimuli, outperforming alternative methods with minimal manual intervention.

This study examines how visual impairment affects brain-computer interface performance using visual spatial attention, finding that while ERPs change, BCI decoding accuracy remains unaffected.

The paper uses resting-state EEG data and a spatial transformer model to classify chronic pain status with an AUROC of 0.782.

AVOCODO integrates behavioral annotation into EEG workflows, improving efficiency and reproducibility in developmental EEG research.

This paper introduces a blind method to identify sleep spindles using trispectral modulation analysis, achieving high agreement with expert annotations and detecting more spindles than conventional methods.

This paper introduces CVIT-ESP and ESPFormer, pre-trained Vision Transformer models adapted for EEG-based seizure prediction, achieving higher AUC-ROC than existing models.

The paper introduces LAND, a framework for improving movement decoding in BCIs by aligning neural and behavioral dynamics using unpaired data, achieving better generalization across movement speeds and subjects.

The paper evaluates EEG foundation models for reliability and disease discrimination, finding they are partially dissociable properties with distinct contributing factors.

This study explores using EEG microstate sequences as BCI triggers for motor imagery classification, achieving 89% accuracy with transfer learning and requiring ~400s for calibration.

BCIJelly provides a unified ecosystem for BCI research, standardizing data, integrating decoders, and enabling efficient deployment on neuromorphic hardware.

This paper validates the Gabor-Nelson method for reconstructing deep-brain oscillatory phase from scalp EEG without anatomical imaging, achieving comparable accuracy to MRI-informed methods.

This paper benchmarks ten deep learning models for MCI and dementia classification using EEG data, finding that compact architectures can be competitive with larger models.

This paper develops a single-channel EEG framework for cross-animal sleep staging in mice, achieving high accuracy and preserving sleep architecture over long-term recordings.

This study uses EEG and machine learning to classify meditation styles and mind-wandering, showing high accuracy within individuals but poor generalization across subjects.

The paper uses EEG data and support vector machines to classify neural responses to words and images, finding higher accuracy for images than words.

This paper introduces a 3D CNN with a trainable CWT layer and BiCL pre-training for sub-second seizure detection, achieving >95% accuracy on EEG data.