EEG and Brain-Computer Interfaces

This paper reconstructs articulatory features from brain signals using generalizable features from able-bodied individuals, achieving high correlation in three participants.

This study evaluates automatic sleep scoring using EEG data from healthy and sleep apnea patients, finding that single-channel scoring is reliable and real-time methods match offline accuracy, though performance drops in clinical populations.

This paper develops a cross-species deep learning model for sleep staging in humans and rodents, achieving high accuracy without retraining.

This paper introduces error modulation in motor BCIs using pre-error motor cortex activity to improve cursor control performance during online BCI tasks.

This paper introduces a wearable BCI system using AI to mitigate car sickness through attention shifting, showing over 83% effectiveness in reducing symptoms.

This paper reviews spatial filters and their limitations in brain data analysis, focusing on virtual electrodes and their practical constraints.

This paper compares the performance of high-channel and low-channel EEG devices in classifying lower limb motor imagery and execution tasks.

This paper develops an automated method to detect quiet and non-quiet sleep in preterm neonates using aEEG, showing strong correlation with brain maturation.

This paper introduces a compositional decoding model for handwriting BCIs that improves generalization by leveraging behavioral motifs.

This study compares the effectiveness of using neural activity from ventral premotor cortex F5c, dorsal premotor cortex, and primary motor cortex for motor brain-machine interface control, finding comparable or better performance from F5c.

CoPrimeEEG reconstructs high-rate EEG from low-rate co-prime samples using a dual-branch neural network with a CRT-guided loss, achieving state-of-the-art quality with fewer parameters.

The paper introduces MANDT, a method that extracts consistent embeddings of motor cortical dynamics to improve BCI decoding across sessions and tasks, validated on rhesus monkey data.

This study compares EEG data quality and reliability between laboratory and clinic settings, finding high agreement in periodic and aperiodic features across age groups.

This paper introduces a co-adaptive decoding method for fMRI neurofeedback that improves training performance through real-time decoder updates, supported by simulations and real data.

This paper introduces ExSEnt, a novel entropy-based measure for dementia detection using EEG, showing it achieves comparable classification performance to deep learning methods with a simpler logistic regression model.

FSINC improves EEG-fMRI source imaging by integrating frequency-resolved EEG with fMRI via a data-driven neurovascular coupling model, achieving better spatial and temporal accuracy in both simulated and real data.

This study evaluates how ensemble learning strategies affect performance and uncertainty estimation in EEG classification under dataset shifts.

G3DCT is a deep learning framework for EEG artifact identification that outperforms existing models, achieving a 2.8% F1-score improvement on combined artifacts.

This study explores using real-time representational similarity analysis in fMRI neurofeedback to navigate emotional states, showing feasibility and improved performance with repeated sessions.

The paper introduces trust-gated augmentation to reduce harm from synthetic data in EEG tasks under subject shift, showing improved generalization and reduced performance degradation.

The paper evaluates how electrode density affects EEG decoding and RSA across three cognitive tasks, finding task-dependent impacts and suggesting low-density setups can be sufficient.

This study decodes cognitive distraction during driving using single-channel in-ear EEG, showing comparable temporal dynamics to full-cap scalp EEG and highlighting eye velocity as an early behavioral marker.

This paper tests an ensemble of sleep staging algorithms on single-channel EEG from a wearable device, achieving high accuracy comparable to expert PSG scoring.

The paper uses time-series models to predict human EEG responses to visual stimuli and finds the simplest AR model performs as well as more complex models.

This paper demonstrates cross-brain transfer learning for intracortical BCIs decoding speech and handwriting, showing improved performance with limited training data and specific decoder design.

This study uses self-supervised learning to classify autism spectrum condition from hyperscanning EEG data during social interactions, achieving 78.13% accuracy.

This paper evaluates auditory attention decoding in conversation-tracking scenarios, finding that accuracy is not affected by attention direction or switches, and correlates with perceived listening effort.

This paper develops a portable C-EPOCS system for real-time neurofeedback, showing that lower-resolution EMG signals can still reliably estimate key parameters for stimulation intensity.

This study investigates how electrode density affects EEG-based neural decoding accuracy using visual and natural stimuli, finding that reliable decoding is possible even with low-density setups.

This study uses dynamic graph neural networks to classify EEG signals and analyze temporal changes in brain connectivity for epilepsy and healthy subject differentiation.