A Human Motor Imagery EEG Dataset Covering Diverse Cognitive States and Neural Response Patterns

The IMU-MI_A dataset is a unique large-scale motor imagery resource supporting cross-paradigm studies of temporal brain dynamics. It includes synchronized EEG and EMG data from 244 participants, encompassing both gross (left/right hand/foot) and fine (individual thumb/index finger movements, pinch gestures) motor tasks. Data are hierarchically structured for batch processing and include demographic details to facilitate model generalizability assessment.Technical validation involved a three-tier approach: (1) Neurophysiological validation confirmed event-related desynchronization/synchronization (ERD/ERS) patterns across participant groups; (2) Classical machine learning using CSP/FBCSP/FTA features with SVM achieved a maximum accuracy of 63.3% ± 1.1%; (3) Deep learning evaluation with EEGNet yielded an average accuracy of 62.4%. These results collectively affirm the dataset's high reliability.With its large sample size, multimodal design, and innovative dual-paradigm approach, IMU-MI_A provides a valuable resource for the BCI and neuroscience communities, advancing the understanding of motor imagery's neural underpinnings and offering robust data support for exploring brain activity patterns across diverse task states.For cooperation or to request more data, please contact : [e-mail] or [e-mail] or [e-mail]