Multi-Electrode ICMS Enables Dexterous Use of Bionic Hands
摘要
To facilitate dexterous object interactions using bionic hands, we need to reliably convey information about contact location and force. Intracortical microstimulation (ICMS) of somatosensory cortex evokes vivid touch percepts where the location and force are determined by electrode and stimulus amplitude, respectively. However, the extent to which these sensations enable functionality in neuroprosthetics is unclear. Three human participants with spinal cord injuries were implanted with four microelectrode arrays, two in primary motor (M1) and two in primary somatosensory (S1) cortex. We delivered ICMS through electrodes on the S1 arrays and asked participants to report the location of the evoked sensation. We used these reports to map sensors on a bionic hand to somatotopically matched electrodes and asked one participant to perform a digit identification task, comparing performance when stimulation was delivered through single or multiple electrodes. An amplitude discrimination task was used to assess the sensitivity of sensory of each participant. One participant also completed a magnitude estimation task to assess the relationship between stimulation amplitude and perceived intensity of both mechanical (on skin) and ICMS (in S1) stimuli. Simultaneous ICMS to electrodes with overlapping projected fields produced additive percepts resembling the superposition of the individual components. Performance in the digit identification task was above chance using single electrode ICMS but improved to near perfect performance when ICMS was delivered through multiple electrodes. By equating the intensity of the ICMS evoked sensations to mechanical ones, we demonstrate that multi-electrode stimulation improves the range of evoked sensations compared to its single-electrode counterpart, allowing for more discrete intensity levels. We show that multi-electrode ICMS evokes more localizable percepts across a wider intensity range which enables ICMS to be effectively used for closed-loop tasks with bionic limbs.