Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Parkinson's Disease Detection from Resting State EEG using Multi-Head Graph Structure Learning with Gradient Weighted Graph Attention Explanations

Christopher Neves, Yong Zeng, Yiming Xiao

TL;DR

This paper tackles Parkinson's disease detection from resting-state EEG by combining Structured Global Convolutions with contrastive self-supervised learning, a dynamic Multi-Head Graph Structure Learner, and a head-wise gradient-weighted graph attention explainer to both improve performance and provide interpretable brain connectivity insights. The proposed MH-GSL learns multiple adjacency matrices in parallel and feeds a Chebyshev-based GNN, while the gradient-weighted explainer highlights task-relevant edges. Evaluations on the UCSD rs-EEG PD dataset show a peak subject-wise LOOCV accuracy of 69.40% with favorable ablations confirming the benefits of each component, and the gradient-weighted explanations reveal more global patterns than stationary graphs. The approach promises clinically meaningful explanations alongside competitive PD detection, with potential extensions to other neurological conditions and multimodal validation through EEG-fMRI.

Abstract

Parkinson's disease (PD) is a debilitating neurodegenerative disease that has severe impacts on an individual's quality of life. Compared with structural and functional MRI-based biomarkers for the disease, electroencephalography (EEG) can provide more accessible alternatives for clinical insights. While deep learning (DL) techniques have provided excellent outcomes, many techniques fail to model spatial information and dynamic brain connectivity, and face challenges in robust feature learning, limited data sizes, and poor explainability. To address these issues, we proposed a novel graph neural network (GNN) technique for explainable PD detection using resting state EEG. Specifically, we employ structured global convolutions with contrastive learning to better model complex features with limited data, a novel multi-head graph structure learner to capture the non-Euclidean structure of EEG data, and a head-wise gradient-weighted graph attention explainer to offer neural connectivity insights. We developed and evaluated our method using the UC San Diego Parkinson's disease EEG dataset, and achieved 69.40% detection accuracy in subject-wise leave-one-out cross-validation while generating intuitive explanations for the learnt graph topology.

Parkinson's Disease Detection from Resting State EEG using Multi-Head Graph Structure Learning with Gradient Weighted Graph Attention Explanations

TL;DR

This paper tackles Parkinson's disease detection from resting-state EEG by combining Structured Global Convolutions with contrastive self-supervised learning, a dynamic Multi-Head Graph Structure Learner, and a head-wise gradient-weighted graph attention explainer to both improve performance and provide interpretable brain connectivity insights. The proposed MH-GSL learns multiple adjacency matrices in parallel and feeds a Chebyshev-based GNN, while the gradient-weighted explainer highlights task-relevant edges. Evaluations on the UCSD rs-EEG PD dataset show a peak subject-wise LOOCV accuracy of 69.40% with favorable ablations confirming the benefits of each component, and the gradient-weighted explanations reveal more global patterns than stationary graphs. The approach promises clinically meaningful explanations alongside competitive PD detection, with potential extensions to other neurological conditions and multimodal validation through EEG-fMRI.

Abstract

Parkinson's disease (PD) is a debilitating neurodegenerative disease that has severe impacts on an individual's quality of life. Compared with structural and functional MRI-based biomarkers for the disease, electroencephalography (EEG) can provide more accessible alternatives for clinical insights. While deep learning (DL) techniques have provided excellent outcomes, many techniques fail to model spatial information and dynamic brain connectivity, and face challenges in robust feature learning, limited data sizes, and poor explainability. To address these issues, we proposed a novel graph neural network (GNN) technique for explainable PD detection using resting state EEG. Specifically, we employ structured global convolutions with contrastive learning to better model complex features with limited data, a novel multi-head graph structure learner to capture the non-Euclidean structure of EEG data, and a head-wise gradient-weighted graph attention explainer to offer neural connectivity insights. We developed and evaluated our method using the UC San Diego Parkinson's disease EEG dataset, and achieved 69.40% detection accuracy in subject-wise leave-one-out cross-validation while generating intuitive explanations for the learnt graph topology.
Paper Structure (14 sections, 2 equations, 2 figures, 1 table)

This paper contains 14 sections, 2 equations, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Works
  3. Methods and Materials
  4. Feature Encoder with Contrastive Learning
  5. Multi-Head Graph Structure Learner
  6. Graph-based EEG Classification
  7. Head-wise Gradient-Weighted Graph Attention Explainer
  8. Dataset and Preprocessing
  9. Experimental Setup and Ablation Studies
  10. Results
  11. Discussion
  12. Conclusion
  13. Acknowledgments.
  14. Disclosure of Interests.

Figures (2)

  • Figure 1: Overview of the model architecture for PD detection.
  • Figure 2: Group-wise mean adjacency matrices for PD and healthy subjects for static PCC, mean head-wise attention, and gradient-weighted mean head-wise attention.