A New Graph-Transformer Framework for EEG-Based Differentiation of Alzheimer’s Disease and Frontotemporal Dementia
Published in Journal of Neural Engineering, 2024
This research presents a novel graph-based transformer architecture that integrates spatial-temporal EEG features for the classification of Alzheimer’s Disease and Frontotemporal Dementia. The work demonstrates state-of-the-art performance in biomedical signal processing applications.
Model Architecture
Graph-based transformer architecture integrating spatial–temporal EEG features
Key Achievements
| Metric | Result |
|---|---|
| Accuracy | 99.1% |
| vs. SOTA | +7-18 percentage points |
- Designed and implemented a novel graph-based transformer integrating spatial–temporal EEG features
- Built an end-to-end processing pipeline and benchmarked classical ML vs. deep learning approaches
- Led a multi-team project, owning the full EEG classification pipeline
EEG Feature Analysis
Power spectral density heatmap comparison showing brain activity differences
Research Details
Role: First author — responsible for design, analysis, and drafting
Code Repository: github.com/yosftag1/eeg-alzheimer-classification
Status: Under Review at IOP Journal of Neural Engineering
Timeline: September 2023 – June 2024 (Bachelor’s Thesis Project)
Preprint: Available upon request