Decoder remapping to counteract neuron loss in brain-machine interfaces
Rodolphe Heliot1, Subramaniam Venkatraman, Jose M Carmena
1CEA-LETI, Minatec, France. rodolphe.heliot@cea.fr
Summary
This study introduces a new brain-machine interface (BMI) method to maintain performance despite neuron loss. The technique adapts filters to compensate for losing up to 24% of neurons with minimal impact on BMI function.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Machine Learning
Background:
- Brain-machine interfaces (BMIs) performance is sensitive to neural signal variability.
- Loss of neurons in the recorded ensemble can degrade BMI control.
- Existing BMI systems often lack robust adaptation to neural degradation.
Purpose of the Study:
- To develop and evaluate a novel adaptive linear filter for BMIs.
- To compensate for performance degradation caused by neuron loss in neural ensembles.
- To minimize performance decline in closed-loop BMI operation.
Main Methods:
- A novel adaptive linear filter technique was developed for BMIs.
- Simulations of closed-loop BMI operation were used to model the learning process.
- The technique's efficacy was assessed under simulated neuron loss conditions.
Main Results:
- The adaptive filter successfully compensated for neuron loss.
- Simulations demonstrated adaptation to the loss of 24% of neurons.
- Performance degradation was limited to 13% despite significant neuron loss.
Conclusions:
- The proposed adaptive filter technique offers a robust solution for maintaining BMI performance.
- This method can mitigate the impact of neural signal variability due to neuron loss.
- The findings suggest improved reliability and longevity for BMI systems.
