Motor and Sensory Control of Robotic Hands Through Fascicular Targeting
Jonathan Cheng, MD, FACS1, Edward Keefer, PhD2.
1University of Texas Southwestern Medical Center, Dallas, TX, USA, 2Nerves Incorporated, Dallas, TX, USA.
PURPOSE: 18-25,000 upper limb amputations occur in the US annually. We have developed fascicular targeting (FAST) as a surgical approach for implanting electrode interfaces within the individual component fascicular groups of the nerves in the residual limb of upper extremity amputees. Our hypothesis is that FAST, when combined with custom-developed technologies including longitudinal intrafascicular electrode (LIFE) interfaces, nerve stimulation and recording electronics, sensory stimulation patterns, and artificial intelligence (AI) motor decoding algorithms, will enable the restoration of naturalistic hand function in robotic hand protheses used by upper extremity amputees.
METHODS: 5 upper extremity amputees were implanted with LIFE + cuff interfaces for durations ranging from 3 months up to 1 year. 3 subjects were partial-hand amputees with 2 FAST interfaces (2 ulnar nerve fascicles). 2 subjects were transradial amputees with 4 FAST interfaces (2 ulnar nerve and 2 median nerve fascicles). Weekly experimental sessions were performed: to assess stimulation parameters for sensory feedback, to record motor signals for robotic hand control, and to train and measure functional performance following restoration of naturalistic sensory feedback and motor control. Implants were removed at the completion of the trial.
RESULTS: All implants were well-tolerated by subjects, with little to no functional morbidity caused by their participation in the trial. Sensory stimulation thresholds remained within a usable range for the duration of the trial. Stimulation patterns were used to elicit sensations perceived as either natural or non-natural, according to subjects' preferences. Sensory discrimination and anatomic localization were also assessed. Motor signals were recorded using a novel microchip design with built-in artifact rejection circuitry. Acquisition of single-unit motor data permitted the use of AI-based signal decoding algorithms to establish independent, free-will control of all 5 digits of the robotic hand. Single-session decodes continued to work for over 3-4 weeks post-training. Functional performance using closed-loop sensorimotor control, with anatomically relevant sensory feedback and user controlled activation of individual digits of the robotic hand, will be discussed in detail.
CONCLUSION: Fascicular targeting, when combined with specialized sensory stimulation and motor recording strategies, can enable true dexterity for upper extremity amputees using robotic hands. This holds promise for the development of full clinical systems to restore the hands of upper extremity amputees.
Back to 2019 Abstracts