…PD, or to build machine learning models for classification. Questions or requests…
Raw 128-channel EEG data from 20 observers viewing object images in a rapid presentation paradigm, together with representational dissimilarity matrices. The dataset supports time-resolved neural decoding analyses that separate objects' contextual associations from their perceptual and conceptual attributes. One participant was excluded because of a technical recording error, leaving 19 usable participants.
…It accompanies Sundrani et al. (2025), "Deep learning on brief interictal intracranial…
…models uniquely predict some part of early neural responses, one based on…
…For each participant, the two machine learning models obtained in the offline…
This dataset contains raw EEG and transcranial magnetic stimulation (TMS) data from healthy adults, collected to test real-time, personalized brain state-dependent stimulation. Whole-brain activity patterns were decoded in real time to predict activation of the corticospinal tract, with motor evoked potentials as the outcome measure. The companion publication (Khatri, Pulliam, Manesiya, Cortez, Millán and Hussain, 'Personalized whole-brain activity patterns predict human corticospinal tract activation in real-time', Brain Stimulation, 2024) is in press and has no DOI yet.
…the P300 speller using language models and various flashing paradigms. For more…
A calibrated task subset of Kai J. Miller's library of human electrocorticographic (ECoG) data, comprising 111 recordings from 24 epilepsy patients with subdural electrode grids. Tasks include finger flexion, hand gestures, motor movement and imagery, and verb generation and noun reading. Data were recorded at 1 kHz and converted to BIDS in BrainVision format with source-documented amplitude calibration.
…The development of subject-independent models eliminates the need to collect data…
A BIDS conversion of a multicenter intracranial EEG (iEEG) dataset of 3-second single-channel clips for classifying graphoelements (physiological and pathological activity) and artifactual signals. Clips were recorded from stereo-EEG contacts at 5000 Hz and concatenated into runs per subject and contact. Events preserve the original labels, segment IDs, anatomy, reviewer IDs, seizure onset zone flags, and electrode types.