A robust nonlinear observer for a class of neural mass models

Area of Science:

• Computational Neuroscience
• Nonlinear Systems Theory
• Biomedical Signal Processing

Background:

• Neural mass models (NMMs) simulate large-scale neural population activity.
• Estimating intracellular neural dynamics like membrane potential from external signals (EEG) is challenging.
• Existing observer designs may lack robustness to noise and model uncertainties.

Purpose of the Study:

• To develop a robust nonlinear observer for NMMs.
• To estimate unmeasured neural membrane potential using electroencephalogram (EEG) data.
• To ensure observer robustness against input uncertainty and measurement noise.

Main Methods:

• Utilized Lur'e system theory for nonlinear system analysis.
• Applied the projection lemma for observer design.
• Designed a robust nonlinear observer tailored for a class of NMMs.
• Validated the observer's performance in estimating membrane potential from simulated EEG.

Main Results:

• Successfully designed a nonlinear observer robust to input uncertainty and measurement noise.
• Demonstrated effective estimation of neural population membrane potential from EEG signals generated by NMMs.
• An illustrative example confirmed the practical efficacy of the proposed observer design.

Conclusions:

• The proposed observer design effectively estimates neural membrane potential from EEG.
• The method offers enhanced robustness for analyzing neural mass models.
• This approach advances non-invasive estimation of neural population dynamics.