Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

NeuroPhysNet: A FitzHugh-Nagumo-Based Physics-Informed Neural Network Framework for Electroencephalograph (EEG) Analysis and Motor Imagery Classification

Zhenyu Xia, Xinlei Huang, Yuantong Gu, Suvash C. Saha

TL;DR

This work tackles EEG motor imagery classification by integrating biophysical neurodynamics into a physics-informed neural network. By embedding the FitzHugh–Nagumo model within a CNN-Transformer PINN architecture, the framework enforces physiologically plausible dynamics, reduces data requirements, and improves robustness to noise and inter-subject variability. Evaluated on BCIC-IV-2a, NeuroPhysNet demonstrates superior generalization and stability under data-limited conditions compared with CSP-based, Riemannian, tensor, and deep-learning baselines, with ablations confirming the value of the derived $v$ (membrane potential) and $w$ (recovery) features. The approach offers a path toward more interpretable, reliable EEG-based diagnostics and assistive technologies in clinical and BCI settings.

Abstract

Electroencephalography (EEG) is extensively employed in medical diagnostics and brain-computer interface (BCI) applications due to its non-invasive nature and high temporal resolution. However, EEG analysis faces significant challenges, including noise, nonstationarity, and inter-subject variability, which hinder its clinical utility. Traditional neural networks often lack integration with biophysical knowledge, limiting their interpretability, robustness, and potential for medical translation. To address these limitations, this study introduces NeuroPhysNet, a novel Physics-Informed Neural Network (PINN) framework tailored for EEG signal analysis and motor imagery classification in medical contexts. NeuroPhysNet incorporates the FitzHugh-Nagumo model, embedding neurodynamical principles to constrain predictions and enhance model robustness. Evaluated on the BCIC-IV-2a dataset, the framework achieved superior accuracy and generalization compared to conventional methods, especially in data-limited and cross-subject scenarios, which are common in clinical settings. By effectively integrating biophysical insights with data-driven techniques, NeuroPhysNet not only advances BCI applications but also holds significant promise for enhancing the precision and reliability of clinical diagnostics, such as motor disorder assessments and neurorehabilitation planning.

NeuroPhysNet: A FitzHugh-Nagumo-Based Physics-Informed Neural Network Framework for Electroencephalograph (EEG) Analysis and Motor Imagery Classification

TL;DR

This work tackles EEG motor imagery classification by integrating biophysical neurodynamics into a physics-informed neural network. By embedding the FitzHugh–Nagumo model within a CNN-Transformer PINN architecture, the framework enforces physiologically plausible dynamics, reduces data requirements, and improves robustness to noise and inter-subject variability. Evaluated on BCIC-IV-2a, NeuroPhysNet demonstrates superior generalization and stability under data-limited conditions compared with CSP-based, Riemannian, tensor, and deep-learning baselines, with ablations confirming the value of the derived (membrane potential) and (recovery) features. The approach offers a path toward more interpretable, reliable EEG-based diagnostics and assistive technologies in clinical and BCI settings.

Abstract

Electroencephalography (EEG) is extensively employed in medical diagnostics and brain-computer interface (BCI) applications due to its non-invasive nature and high temporal resolution. However, EEG analysis faces significant challenges, including noise, nonstationarity, and inter-subject variability, which hinder its clinical utility. Traditional neural networks often lack integration with biophysical knowledge, limiting their interpretability, robustness, and potential for medical translation. To address these limitations, this study introduces NeuroPhysNet, a novel Physics-Informed Neural Network (PINN) framework tailored for EEG signal analysis and motor imagery classification in medical contexts. NeuroPhysNet incorporates the FitzHugh-Nagumo model, embedding neurodynamical principles to constrain predictions and enhance model robustness. Evaluated on the BCIC-IV-2a dataset, the framework achieved superior accuracy and generalization compared to conventional methods, especially in data-limited and cross-subject scenarios, which are common in clinical settings. By effectively integrating biophysical insights with data-driven techniques, NeuroPhysNet not only advances BCI applications but also holds significant promise for enhancing the precision and reliability of clinical diagnostics, such as motor disorder assessments and neurorehabilitation planning.

Paper Structure

This paper contains 27 sections, 29 equations, 1 figure, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. FitzHugh-Nagumo Model
  5. Design and Implementation of a Physics-Informed Neural Network Based on the FitzHugh-Nagumo Model
  6. Data Preprocessing Module
  7. Physics-Informed Neural Network Module
  8. Input Processing and Convolutional Layers
  9. Fully Connected Layer
  10. Transformer Encoder for Temporal Dependencies
  11. Output Layer and Reshaping
  12. Design Rationale
  13. physics-based Loss Function
  14. Coupling Term
  15. Final Physics Loss
  16. ...and 12 more sections

Figures (1)

  • Figure 1: Visualization of $v$ (membrane potential) during the Wake stage and $w$ (recovery variable) during the REM stage for selected EEG channels.