Wavelet Entropy-Based Inter-subject Associative Cortical Source Localization for Sensorimotor BCI

Published on July 23, 2019

We propose event-related cortical sources estimation from subject-independent electroencephalogram (EEG) recordings for motor imagery brain computer interface (BCI). By using wavelet-based maximum entropy on the mean (wMEM), task-specific EEG channels are selected to predict right hand and right foot sensorimotor tasks, employing spatial pattern analysis with and without covariance estimation regularization. EEG from five healthy individuals (Dataset IVa, BCI Competition III) were evaluated by a cross-subject paradigm. Prediction performance was evaluated via a two-layer feed-forward neural network, where the classifier was trained and tested by data from two subjects independently. The highest mean prediction accuracy achieved by using subject-pair (ay-al) specific selected EEG channels was on average (90:36+/-5:59) and outperformed that achieved by using all available channels (83:21+/-12:26). Significant improvements in performance suggest a role of wMEM for channel selection in BCI. Spatially projected cortical sources may be useful for capturing inter-subject associative sensorimotor brain dynamics and pave the way towards subject-independent BCI.

Read Full Article (External Site)