Towards Implementation of Motor Imagery using Brain Connectivity Features

This study aims to explore modulation of the connectivity pattern when people per-form left hand versus right hand motor imagery and probe the feasibility of adopting connectivity information to discriminate these tasks. Nine subjects were recorded with 16-channel EEG system, covering sensorimotor cortex. Non-normalized directed transfer function (DTF) is used to obtain the brain connectivity between EEG electrodes. The results demonstrate that the modulations of intrahemispheric and interhemispheric infor-mation flows are not identical during left and right hand motor imageries. Particularly, the mu rhythm is highly modulated in intrahemispheric brain interactions, whereas the high frequency bands are more related with distant interhemispheric brain interactions. Fur-thermore, classification results suggest that the DTF features bring additional informative features for the classification between two tasks.