Visualization of a functional visual cognition network by electrocorticogram

Abstract

In order to better interpret spatial and temporal changes on electrocorticograms (ECoG) taken during semantic tasks, we developed software to visualize semantic-ECoG dynamics on individual brains. Twenty patients with intractable epilepsy underwent implantation of subdural electrodes (more than 80 channels) bilaterally. Semantic-ECoGs were then recorded during word, figure and face recognition tasks. The ECoG raw data was processed by averaging and time-frequency analysis and the functional profiles were projected on individual brain surfaces. Acquired ECoG was classified by Support Vector Machine and Sparse Logistic Regression to classify brain signals evoked by different stimuli. Because of electrode location variations, we normalized the ECoG elecrtrodes using SPM8. The basal temporal-occipital cortex was activated within 250 msec after visual object presentations. Face stimulation evoked significantly higher ECoG amplitudes than other stimuli. The prediction rate of ECoG-classification reached 90, which is sufficient for clinical use. Semantic-ECoG is a powerful technique to detect and decode human brain functions.