EEG and Brain-Computer Interfaces

This paper introduces a self-supervised transformer model for sleep-EEG data that outperforms supervised methods and can be fine-tuned for new tasks like individual identification and brain age estimation.

This paper uses a Switching State Space Model to automatically detect high-frequency oscillations in intracranial EEG data, improving accuracy compared to manual annotation.

This study evaluates six EEG data augmentation methods for MDD diagnosis, finding that most do not improve performance beyond using duplicate data, but channel dropout augmentation does.

This paper introduces myelin-informed forward models for M/EEG source reconstruction, showing improved accuracy using myelin density estimates from MRI.

This paper shows that CSF-filled cavities significantly affect scalp EEG amplitude and topography, making group comparisons difficult to interpret without accounting for this confounding factor.

This paper shows that targeting V5 with transcranial focused ultrasound improves a visual motion BCI task by enhancing feature-based attention through increased theta activity and preserved dorsal pathway connections.

This paper decodes absolute auditory attention using EEG features, finding that neural envelope tracking is more reliable than spectral entropy for distinguishing active listening from distractor conditions.

This paper introduces BrainState, a self-supervised method to model complex brain states from EEG data, applied to epilepsy patients to capture pre-seizure states and neuromodulation effects.

This study improves EEG-based MDD diagnosis using transfer learning by adapting representations from a sleep dataset to a higher sampling rate, increasing model accuracy and robustness.

The paper tests two high-density EEG sensor net designs in infants with curly or tightly-coiled hair, finding that the tall design improves data quality and retention.

This study evaluates methods for estimating low-frequency steady-state evoked potentials (SSEP) using short data recordings and identifies optimal parameters for reliable detection.

The paper develops a deep learning model to remove EOG artifacts from EEG data using eye-tracking data, showing that ICA has low specificity for EOG removal.

This paper decodes speech features from EEG signals using deep learning models, achieving above-chance performance for both discrete and continuous speech tasks.

The paper develops a computational framework using control and game theory to predict and optimize user-decoder interactions in co-adaptive neural interfaces through experimental validation.

This study uses a Transformer model trained on overt speech to decode covert speech from ECoG data, achieving similar performance to models trained directly on covert speech.

The paper compares dipole source reconstruction accuracy between 3-layer and 5-layer BEM-FMM models, finding that 3-layer models provide good localization but less accurate orientation, with AMR improving 5-layer accuracy.

This paper introduces a deep neural network framework that directly predicts the complex-valued analytic signal of EEG from raw data, outperforming traditional methods and showing robustness to missing electrodes.

The paper evaluates denoising techniques for EEG data during transcutaneous auricular vagus nerve stimulation, finding that GED/SASS and SSP are most effective in reducing artifacts while preserving signal integrity.

This paper combines gaze tracking with the Stimulus-Preceding Negativity (SPN) to improve target selection in XR by confirming user intent implicitly.

PyHFO is a lightweight deep learning platform for efficient HFO detection in EEG data, achieving 50x speed improvements over standard methods.

This study uses Explainable AI to analyze uhd-EEG signals for speech decoding, showing neural contributions are distinct from EMG artifacts and can be enhanced by denoising.

This paper conducts a systematic review and meta-analysis of neurofeedback studies on frontal midline theta modulation in healthy humans, finding a medium effect size with high heterogeneity.

This study develops an EEG neurofeedback-guided cognitive reappraisal training protocol for emotion regulation, showing improved performance for low positive valence stimuli with real-time feedback.

The paper classifies working memory loads using Bayesian functional connectivity and graph convolutional networks, achieving 96% accuracy in intrasubject classification.

This paper introduces a convolutional neural network for accurate M/EEG source localization, validated on simulations and real data, showing improved accuracy for both cortical and subcortical regions.

This paper reviews Neurourbanism studies using mobile EEG, identifies methodological flaws, and provides guidelines for better practice.

This paper develops a flexible intracortical BCI for typing using finger movements, achieving high accuracy and speed in two typing paradigms.

This study shows that during BMI learning, animals transition from a global strategy involving all neurons to a local strategy involving only direct neurons as they become experts.

This paper develops a theory explaining BCI learning through low-dimensional control strategies, validated with experimental data.

This paper introduces a method to infer brain activity sources and their locations from electromagnetic signals using linear analysis and anatomical models.