514 A Functional Ultrasound Brain-Machine Interface: Real-Time Decoding of Eight Movement Directions. (April 2023)

• Record Type: Journal Article
• Authors: Griggs, Whitney; Norman, Sumner; Rabut, Claire; Deffieux, Thomas; Christopoulos, Vasileios; Liu, Charles Y.; Tanter, Mickael; Shapiro, Mikhail; Andersen, Richard
• Is Part Of: Neurosurgery. Volume 69(2023)Supplement 1
• Page Start: 114
• Page End: 115
• Publication Date: 2023-04
• Subjects: Nervous system -- Surgery -- Periodicals; 617.48005
• DOI: 10.1227/neu.0000000000002375_514
• Languages: English
• ISSNs: 0148-396X

Abstract

Abstract : INTRODUCTION: Brain-machine interfaces (BMIs) can be transformative for people living with chronic paralysis. By translating brain signals into computer commands, BMIs can bypass neurological impairments, enabling users to control computers, robots, and more. State-of-the-art BMIs have already made this future a reality in limited clinical trials. However, these BMIs requires inserting highly invasive electrodes into the brain. Device degradation limits longevity to 3-5 years and the field of view of brain activity is small. These factors limit BMI's widespread adoption. The next generation of BMIs must be longer lasting, less invasive, and scalable. Functional ultrasound neuroimaging (fUSI) is a recently developed technique that meets these criteria. METHODS: We collected data from two nonhuman primates (NHPs). We used real-time fUSI to measure activity in the posterior parietal cortex, an important area for motor planning. We fed this fUSI data into real-time machine learning classifiers to predict the monkey's intended movement direction and control a behavioral task in real-time. RESULTS: We could predict the monkey's intended movement direction significantly above chance level (binomial test, p<0.001 for 2- and 8-directions) using the real-time fUSI-BMI. Most direction misclassifications were for neighboring directions (permutation test, p<0.001). The most important voxels for decoding eye movements were within small subpopulations of the lateral intraparietal area (LIP). We found that these LIP subpopulations had a stable encoding of movement direction and the NHPs could accurately control the task more than 40 days after training the decoder model. CONCLUSIONS: This work establishes, for the first time, the feasibility of a real-time ultrasound-based BMI and we have begun translating this work into human applications. This work contributes to the next generation of high-performing, yet minimally-invasive, BMIs that can restore function to patients with neurological impairments.