THE 5TH INTERNATIONAL CONFERENCE BCI: SCIENCE AND PRACTICE. SAMARA 2019 ECOG based inverse modelling for decoding and eloquent cortex mapping
Solving the ECoG inverse problem reconstructs neural source activity , which may improve decoding in neural prostheses and can be used for accurate delineation of eloquent cortex in neurosurgical planning.
ECoG data was recorded from 8x8 ECoG grid located over the left sensorimotor cortex of a patient with epilepsy conducting flexion-extension motions of right-hand digits for 1 minute. Digit trajectories were by Perception Neuron . Anatomical model was based on structural MRI and CT. Lead-field matrix was computed with OpenMEEG . Five approaches of solving the inverse problem were implemented in MATLAB : MNE, wMNE, sLORETA, eLORETA, and LCMV beamformer [5, 6, 7]. Decoding was done using instantaneous power values in 8 narrow bands of the sensor or source-reconstructed ECoG data. Data was divided into training and test sets, Pearson correlation …