(2) An Automated Approach to Detect Persistent Sensations Caused by Intracortical Microstimulation

Introduction: Intracortical microstimulation (ICMS) in the somatosensory cortex (S1) can be used in closed-loop brain computer interfaces (BCIs) to provide artificial somatosensory feedback and can improve motor BCI performance. When a tactile sensation is elicited using ICMS, it is expected to last exactly as long as the stimulation. However, on very rare occasions (41 cases in over 260,000 trials), some participants in our studies report sensations that outlast the stimulation period, typically by several seconds. The cause and underlying neurophysiological mechanisms of these persistent sensations are unknown.

Methods: Three people were implanted with microelectrode arrays in tactile regions of S1 as part of a clinical trial of an intracortical sensorimotor BCI (NCT01894802) under an FDA Investigational Device Exemption. The Neuroport System (Blackrock Neurotech) allows us to simultaneously stimulate and record from these microelectrode arrays.

Results: Two participants reported percepts that outlasted stimulation at least once throughout their participation. In some of these cases, we observed a large amplitude signal resembling self-limiting afterdischarges. Apart from the extended percept duration, the only other effects the participants reported were infrequent and slight changes in the quality of the sensation. The sensations remained localized to the region of the hand where stimulation initially evoked a percept. The neural activity during these events occurred across all channels on the array containing the stimulation electrode, with the maximum amplitude (1.58 ± 0.23 mV) centered on or near the stimulated electrode. The activity did not occur on any other array in the motor cortex or S1, demonstrating highly localized activity. In all cases, the activity vanished within 20 seconds after stimulation ended (12 ± 4.3s). To further investigate this phenomenon, we developed an automated detection algorithm to continuously monitor neural activity during experiments to document the frequency and duration of these events regardless of the presence of a conscious perceptual experience.

Conclusion: The extended percept duration that sometimes occurs is undesirable for accurate somatosensory feedback in a BCI context, and the potential effects of the self-limiting afterdischarges are unclear. Future work is needed to determine the conditions that cause these perceptual changes, which can lead to a set of parameter boundaries that provide reliable and temporally accurate ICMS.