Neural Control of Individual Finger Movements Using Intracortical Recordings From a Person With Tetraplegia. (1st September 2019)

• Record Type: Journal Article
• Authors: Jorge, Ahmed; Royston, Dylan; Tyler-Kabara, Elizabeth C; Boninger, Michael; Collinger, Jennifer
• Is Part Of: Neurosurgery. Volume 66(2010)Supplement 1
• Publication Date: 2019-09-01
• Subjects: Nervous system -- Surgery -- Periodicals; 617.48005
• DOI: 10.1093/neuros/nyz310_209
• Languages: English
• ISSNs: 0148-396X

Abstract

Abstract: INTRODUCTION: Intracortical microelectrode arrays have allowed people to control robotic arm movements such as reaching and grasping. However, to restore dexterous movement, control of fingers will be required. We aimed to control individual finger movements of a virtual-reality hand on a person with tetraplegia. METHODS: A 31-yr-old man with a C5/6 ASIA B spinal cord injury was implanted with two 88-channel intracortical microelectrode arrays (4 × 4 mm footprint, 1.5 mm shank-length) in the left motor cortex. Across 4 d, a 6-class linear discriminant classifier was used as an online decoder that output finger velocity commands every 20 ms. We quantified the number of channels and location significantly modulating each attempted finger movement using a Kolmogorov-Smirnov test. We also report the success rate to reach targets for flexing each of the five fingers and thumb abduction. RESULTS: On average, there were 28 channels modulated by attempted finger movement on the lateral array as compared to 19 on the medial array. Attempted thumb flexion exhibited the highest (n = 18, P < .05) while ring finger had the fewest (n = 13, P < .05) modulated channels. The mean success rate was 61 ± 15% (chance: 17%). The participant was successful for 81% of the thumb flexion trials, while thumb abduction, index, middle, ring and pinky flexion achieved average accuracies of 75%, 63%, 49%, 38%, and 56% respectively. Of the failed trials, 93% failed due to co-activation of adjacent fingers, with middle and ring being the most interdependent. For example, the ring finger successfully flexed on 93% of the "ring finger" trials (high sensitivity), but middle was co-activated during 53% of the same trials (low specificity) resulting in trial failure. CONCLUSION: A person with tetraplegia was able to use a brain-computer interface to control individual digits of a virtual robotic hand. Failed trials typically resulted from movement of adjacent fingers. This co-dependence should be accounted for in future control schemes.