Related Experiment Video
Updated: Jan 17, 2026
09:47
Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches
Published on: December 15, 2023
1.7K
Olfactory EEG based Alzheimer disease classification through transformer based feature fusion with tunable Q-factor
Berke Cansiz1, Hamza Osman Ilhan2, Nizamettin Aydin3
1Department of Electronics and Communication Engineering, Yildiz Technical University, Istanbul, Türkiye.
Frontiers in Neuroscience
|September 15, 2025
Summary
Deep learning models analyzing electroencephalography (EEG) signals from olfactory memory responses achieved 93.14% accuracy in detecting Alzheimer's disease. This novel approach offers a promising, cost-effective diagnostic tool.
Area of Science:
- Neuroscience
- Artificial Intelligence
- Medical Diagnostics
Background:
- Alzheimer's disease (AD) is a leading neurodegenerative disorder characterized by memory loss, significantly impacting daily life.
- Early diagnosis of AD is critical for effective management but current methods like imaging are costly and require expert interpretation.
- Deep learning (DL) presents a potential solution for developing advanced decision support tools for AD detection.
Purpose of the Study:
- To investigate the efficacy of a DL model in detecting Alzheimer's disease using electroencephalography (EEG) signals.
- To leverage olfactory memory responses as a novel biomarker for AD detection.
- To develop a classification model integrating multiple feature types for enhanced diagnostic accuracy.
Main Methods:
- A dataset comprising healthy individuals, patients with amnestic mild cognitive impairment, and Alzheimer's disease patients was utilized.
- A transformer-based fusion approach was employed to integrate three distinct feature types derived from EEG signals.
- Feature vectors were extracted using Common Spatial Pattern, Covariance matrix-Tangent Space, and a Tunable Q-Factor wavelet coefficient mapping.
Main Results:
- The proposed DL model achieved a subject-based classification accuracy of 93.14% for Alzheimer's disease detection.
- Classification was based on EEG-recorded olfactory memory responses to rose aroma.
- The results indicate a high degree of accuracy in differentiating between healthy individuals and AD patients.
Conclusions:
- The study successfully demonstrated the potential of using DL models with EEG-based olfactory memory responses for accurate Alzheimer's disease detection.
- This approach shows superiority compared to existing EEG-based methods reported in the literature.
- The findings suggest a promising, potentially more accessible, diagnostic avenue for Alzheimer's disease.
