From monkeys to humans: observation-basedEMGbrain-computer interface decoders for humans with paralysis
Fabio Rizzoglio1, Ege Altan1,2, Xuan Ma1
1Department of Neuroscience, Northwestern University, Chicago, IL, United States of America.
Journal of Neural Engineering
|October 16, 2023
Summary
Researchers developed new methods for brain-computer interfaces (BCIs) to predict muscle activity by using data from another individual. This advance could improve prosthetic control for people with paralysis.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Rehabilitation Technology
Background:
- Intracortical brain-computer interfaces (iBCIs) typically use observed movements for decoder training, limiting their use for predicting unobservable outputs like muscle activity.
- Existing iBCI methods cannot directly decode intended muscle activity due to paralysis preventing direct training.
- There is a need for novel iBCI decoding strategies to enable control of advanced prosthetics and functional electrical stimulation.
Purpose of the Study:
- To investigate the feasibility of using muscle activity recordings from a surrogate individual to train iBCI decoders.
- To determine if neural representations of motor behavior are conserved across individuals and species for decoder transfer.
- To develop iBCI decoding methods capable of predicting electromyographic (EMG) signals.
Main Methods:
- Two decoding approaches were tested using neural data from a human iBCI user and a monkey performing similar tasks.
- A direct decoding approach trained a model to predict monkey EMG from human neural signals.
- A transfer decoding approach aligned latent neural signals between species using Canonical Correlation Analysis to enable decoder transfer.
Main Results:
- Both direct and transfer decoding methods successfully predicted EMG activity with high accuracy.
- Accurate EMG predictions were achieved between two monkeys and from a monkey to a human.
- The study demonstrated the effectiveness of cross-species and cross-individual decoder transfer.
Conclusions:
- Latent neural representations of motor behavior are consistent across individuals and even across primate species.
- These findings represent a significant step towards developing iBCI decoders that predict EMG signals for prosthetic control.
- The developed methods could enable biomimetic control of prosthetic arms and functional electrical stimulation for restoring movement and impedance.
