Comparative analysis of system identification techniques for nonlinear modeling of the neuron-microelectrode junction

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computational Modeling

Background:

• Non-invasive neuroelectronic interfacing relies on accurate neuron-electrode junction signal processing.
• Existing linear models inadequately represent extracellular signals, limiting applications in biosensing and neural prosthetics.

Purpose of the Study:

• To develop computationally inexpensive and efficient mathematical models for the neuron-electrode junction.
• To evaluate the performance of parametric Wiener and Nonlinear Auto-Regressive network with eXogenous input (NARX-NN) models.

Main Methods:

• Modeled neuron-electrode junction input-output data using parametric Wiener and NARX-NN models.
• Validated model performance against a dataset.
• Compared computational complexity and efficiency with the Lee-Schetzen cross-correlation technique.

Main Results:

• Parametric Wiener and NARX-NN models demonstrated potential for efficient neuron-electrode junction modeling.
• These models offer improved representation of extracellular signals compared to linear circuit models.
• Comparative analysis highlighted the computational advantages of the proposed models.

Conclusions:

• Parametric Wiener and NARX-NN models provide viable, computationally efficient alternatives for modeling the neuron-electrode junction.
• Accurate modeling is crucial for advancing neuroelectronic interfacing applications.
• Further research can optimize these models for enhanced neuroprosthetic and biosensing devices.