EEG and Brain-Computer Interfaces

This paper reconstructs photorealistic texture images from EEG signals using an MVAE-based approach, showing that texture perception can be decoded from low-resolution neural data.

This paper introduces brMEGA, an algorithm for automated detection and removal of cardiogenic artifacts in single-channel EEG, showing effective artifact removal and potential for heart rate feature extraction.

This study classifies sleep stages using multimodal electrophysiology data and introduces an ablation method to assess modality importance, finding EEG is most important for most stages and EOG for REM and nonREM.

This study develops a brain-machine interface for virtual self-motion by decoding neural activity from monkeys performing non-reaching arm movements, showing that different movement classes require distinct decoding strategies.

The paper adapts an adult ICA component classifier for infant EEG data, showing improved performance in artifact removal and classification accuracy compared to both the original adult classifier and manual methods.

This paper introduces an automated sleep staging tool with high accuracy, validated on diverse populations, and provides an open-source, free, and user-friendly solution.

This paper introduces a Kalman filter-based method for real-time estimation of EEG oscillatory phase to enable closed-loop sensory stimulation, showing higher accuracy than autoregressive models.

This study examines multiscale entropy (MSE) in clinical-EEG data to assess its ability to distinguish health status across neurological disorders and non-brain comorbidities.

This study shows humans can volitionally control single neuron firing rates in memory-related brain structures through neuroprosthetic training, with associated changes in neural activity patterns.

CluSem improves MI-EEG classification accuracy by 5-15% using a clustering-based ensemble method and is evaluated in a brain game called CluGame.

This study uses fMRI spatial priors in a Bayesian framework to improve MEG source reconstruction accuracy using simulated and experimental data.

This study investigates alpha activity neuromodulation through neurofeedback training in a double-blind randomized study, finding increased alpha activity only in the neurofeedback group.

The paper compares EEG spectral power changes in patients with opioid-use disorder to those with methamphetamine and alcohol-use disorders, finding significant differences across multiple analytical perspectives.

The paper tests covariance models for MEG source reconstruction, finding that incorporating a correlated hippocampal source model improves model evidence.

The paper analyzes time series variability of MEG power spectra during different cognitive states and finds that dominant frequency spectra vary as weakly stationary processes.

This paper introduces data-driven neural networks with linear-phase filters for detecting epileptic spikes in EEG data, achieving an average AUC of 0.967.

This paper develops a dynamic selective auditory attention detection system using RNN and reinforcement learning, achieving 94.2% classification accuracy with minimal delay.

This paper proposes a state-space model for estimating brain effective connectivity from EEG signals using Granger causality, showing improved performance over existing methods in simulated and real data.

This paper introduces a soft channel attention mechanism for auditory attention detection using EEG signals, achieving high accuracy across different channel counts and decision window sizes.

This study shows that heart rate variability predicts decline in BCI control performance during voluntary sensorimotor rhythm control in healthy participants.

This paper unifies three sparse Bayesian learning algorithms for EEG/MEG brain source imaging under the majorization-minimization framework and introduces a new method for low SNR settings and noise variance estimation.

This paper presents a framework using EEG signals to classify video game player expertise levels, achieving 98.04% accuracy with K-nearest neighbor classifier.

ConvDip is a convolutional neural network that solves the EEG inverse problem using a distributed dipole model, outperforming existing methods in accuracy and efficiency.

This paper introduces a data-driven beamforming technique to reduce BCG artefacts in EEG-fMRI without using ECG signals, improving signal recovery for task-based neural activity.

This paper develops an artifact rejection method for MEG data during speech production, showing it effectively removes speech-related distortions.

This study decodes gait phase information during motor imagery using EEG and shows that combining motor imagery with action observation improves decoding accuracy.

This study shows deep learning improves BCI classification accuracy and information transfer rate using full scalp EEG data, benefiting users with lower initial performance.

This paper develops a deep learning algorithm to detect and classify artefacts in neonatal EEG signals, achieving high accuracy in identifying artefact types and clean EEG periods.

This paper demonstrates a brain-computer interface that decodes imagined handwriting movements into text at high speed and accuracy, closing the gap between BCI and able-bodied typing rates.

This paper introduces a novel method to remove muscle artefacts from EEG data, improving gamma band signal quality by preserving task-relevant cortical activity.