Computational model of neuronal recruitment during ICMS for restoring somatosensation in the human somatosensory cortex

ObjectiveIntuitively providing touch feedback from artificial hands to users with sensory loss remains a challenge. Although localized fingertip sensations can be evoked via intracortical microstimulation (ICMS), feedback is generally optimized using psychometric tasks rather than mimicking the cortical response to touch. ApproachWe created an anatomically-informed and participant-specific model of the human somatosensory cortex (S1) region with an implanted microelectrode array (MEA). We performed simultaneous stimulation-and-recording from the study participant S1 region to characterize cortical responses elicited by single ICMS pulses. Pulses were delivered to a set of pre-selected electrodes mapped to tactile receptive fields. We next performed a 2D (i.e., in the plane of the MEA probe tips) current source density (CSD) analysis of recorded cortical responses to inform cortical network model parameters on how ICMS activates neurons and lateral synaptic connections in the area of the S1 sampled by MEA electrodes. Using information from planar CSD profiles obtained from ground truth data, we reconstructed lateral connections in the S1 model needed to produce the desired responses to single ICMS pulses. The effect of multiple ICMS was then simulated in the biologically realistic cortical model and the results were validated against ground truth cortical responses from the study participant. Main resultsA high-resolution cortical network model, calibrated to produce the known cortical responses to single ICMS pulses delivered to individual electrodes, predicted with a reasonable accuracy the cortical response to ICMS pulses delivered simultaneously to multiple electrodes. SignificanceThese preliminary results suggest that high-resolution biologically realistic cortical network models can potentially be reliable predictors of cortical response to a given pattern of ICMS presentations and therefore useful in designing biomimetic stimulation patterns.