Related Experiment Video
Updated: May 12, 2026
11:25
Simultaneous Scalp Electroencephalography (EEG), Electromyography (EMG), and Whole-body Segmental Inertial Recording for Multi-modal Neural Decoding
Published on: July 26, 2013
Design and validation of a real-time spiking-neural-network decoder for brain-machine interfaces
Julie Dethier1, Paul Nuyujukian, Stephen I Ryu
1Department of Bioengineering, Stanford University, Stanford, CA 94305, USA.
Journal of Neural Engineering
|April 12, 2013
Summary
Spiking neural networks (SNNs) can decode neural signals for brain-machine interfaces (BMIs), matching traditional methods. This demonstrates the potential for low-power, implanted prostheses using neuromorphic chips.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Computer Science
Background:
- Cortically-controlled motor prostheses aim to restore lost function due to neurological conditions.
- Clinical translation of intracortical prostheses is hindered by power dissipation constraints.
- Ultra-low power neuromorphic chips offer a potential solution for decoding neural signals in implants.
Purpose of the Study:
- To explore the feasibility of translating brain-machine interface (BMI) decoding algorithms into spiking neural networks (SNNs) via simulation.
- To validate the use of SNNs for neural signal decoding in intracortical prostheses.
Main Methods:
- Implemented a Kalman-filter-based decoder within a simulated SNN using the Neural Engineering Framework (NEF).
- Tested the SNN decoder's robustness and generalization in closed-loop BMI experiments with two rhesus monkeys.
Main Results:
- The SNN-implemented Kalman filter demonstrated performance comparable to standard floating-point implementations.
- The system showed robustness and generalization across different monkeys and tasks.
Conclusions:
- SNNs are a viable platform for implementing statistical signal processing algorithms for BMIs.
- Neuromorphic chip-based SNN decoders show promise for low-power, fully-implanted prostheses.
- The validated closed-loop decoder system supports the potential of SNNs for advanced neuroprosthetics.
