Inferring the stability of LIFE through Brain Machine Interfaces

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Chronic implantation of neural electrodes is crucial for long-term Brain Machine Interface (BMI) studies.
• Maintaining signal quality and stability from implanted electrodes is a significant challenge.

Purpose of the Study:

• To evaluate the performance of Longitudinally implanted Intra-Fascicular Electrodes (LIFE) in a chronic rabbit model.
• To assess the impact of signal processing techniques on neural signal quality and spike sorting accuracy.
• To investigate how unit discrimination affects Brain Machine Interface (BMI) decoding performance and electrode stability.

Main Methods:

• Utilized translation-invariant wavelet de-noising to enhance signal-to-noise ratio (SNR) of neural signals.
• Applied template-based spike sorting to discriminate individual neural units.
• Assessed Brain Machine Interface (BMI) decoding performance using sorted neural units.
• Evaluated the stability of LIFE by analyzing decoding performance over time and with adaptive BMI methods.

Main Results:

• Wavelet de-noising successfully improved the signal-to-noise ratio (SNR) of neural recordings.
• Template-based spike sorting effectively discriminated single neural units from the LIFE recordings.
• Discriminating between identified units positively impacted Brain Machine Interface (BMI) decoding performance.
• Decoding performance analysis provided insights into the stability of LIFE, even with signal degradation.

Conclusions:

• Longitudinally implanted Intra-Fascicular Electrodes (LIFE) show potential for chronic neural recording.
• Advanced signal processing techniques, including wavelet de-noising and spike sorting, are vital for maximizing data quality.
• The study demonstrates a method to assess and infer the long-term stability of neural implants for BMI applications.