Comparison of electrical microstimulation artifact removal methods for high-channel-count prostheses

Area of Science:

• Neuroscience
• Biomedical Engineering

Background:

• Neuroprostheses stimulate the brain to restore function but generate artifacts that obscure neural signals.
• Validating artifact removal methods is challenging due to contaminated ground-truth data.

Purpose of the Study:

• To evaluate software-based artifact removal methods for high-channel-count cortical visual prostheses (CVP).
• To create a simulated dataset for method validation.

Main Methods:

• Delivered stimulation to the visual cortex via a CVP, recording neural activity and artifacts.
• Quantified artifact properties to simulate data with neuronal activity and artifacts.
• Evaluated six artifact removal algorithms: Template subtraction, Linear interpolation, Polynomial fitting, Exponential fitting, SALPA, and ERAASR.

Main Results:

• Polynomial fitting and Exponential fitting excelled at recovering spikes and multi-unit activity (MUA).
• Linear interpolation and Template subtraction effectively recovered local-field potentials.
• Simulated data facilitated performance evaluation of artifact removal techniques.

Conclusions:

• Polynomial fitting and Exponential fitting offer an optimal balance between signal recovery quality and computational complexity for CVPs.
• These methods are promising for improving neural signal analysis in CVP applications.