A Biomimetic Approach Towards a Dexterous Neuroprosthesis

PROJECT SUMMARY
Cervical spinal cord injury results in the loss of arm and hand function, which significantly limits independence
and results in costs over the person’s lifespan. A brain-computer interface (BCI) can be used to bypass the
injured tissue to enable control of a robotic arm and to provide somatosensory feedback. Two primary limitations
of current state-of-the-art BCIs for arm and hand control are: (1) the inability to control the forces exerted by the
prosthetic hand and (2) the lack of somatosensory feedback from the hand. In the proposed study, we seek to
considerably improve dexterous control of prosthetic limbs by implementing decoding strategies that enable the
user to not only control the movements of the arm and hand, but also the forces transmitted through the hand.
We anticipate that our biomimetic approach to decoding will yield intuitive, dexterous control of the prosthetic
hand. Tactile sensations will be conveyed to the user through intracortical microstimulation (ICMS) of
somatosensory cortex. The spatiotemporal patterns of stimulation will be based on our basic scientific
understanding of how tactile information is encoded in somatosensory cortex, which we expect will result in more
natural and intuitive sensations. In order to achieve our goal of developing a dexterous neuroprosthesis, we have
brought together a team with human BCI experience from the University of Pittsburgh along with the basic
science expertise at both Pitt and the University of Chicago. We will collaborate with experts in implantable
neurotechnology (Blackrock Microsystems) and robotics (The Biorobotics Institute) to ensure that the device
hardware allows us to take a biomimetic approach for control and feedback with an eye toward clinical translation.
A total of 4 participants will be tested in a multisite study to accomplish the following three specific aims. Aim 1:
Evoke natural and intuitive tactile sensations through ICMS of somatosensory cortex. We expect that biomimetic
ICMS will evoke sensations that more closely resemble everyday tactile sensations and intuitively convey
information about contacted objects than does standard fixed-frequency ICMS. Aim 2: Derive kinematic and
kinetic signals from motor cortex for hand control. We will assess the degree to which motor cortical neurons
encode forces exerted on objects. Based on these observations, we will develop hybrid decoders that enable
controlling both the movement and force using a synergy-based approach. Aim 3: Demonstrate improved arm
and hand function with a biomimetic sensorimotor BCI that combines the sensory feedback developed in Aim 1
with the hybrid decoding developed in Aim 2. A battery of functional assessments will be used including novel
metrics designed specifically for sensorimotor prosthetics along with well-established tests identified in the NIH
Common Data Elements. We anticipate that subjects will substantially improve their dexterity using a biomimetic
BCI as compared to non-biomimetic BCIs or BCIs without somatosensory feedback.

Grant Number: 5UH3NS107714-05
NIH Institute/Center: NIH
Principal Investigator: MICHAEL BONINGER