Entropy and Coherence Features in EEG-Based Classification for Alzheimer's Disease Detection

Alzheimer's disease (AD) is a progressive neurode-generative condition that impacts cognitive functions and the overall quality of life of millions of individuals worldwide. Early and accurate diagnosis of AD is crucial for providing appropriate treatment. In this paper, an algorithm that uses...

Full description

Saved in:
Bibliographic Details
Main Authors: Criscuolo, Sabatina, Cataldo, Andrea, De Benedetto, Egidio, Masciullo, Antonio, Pesola, Marisa, Schiavoni, Raissa
Format: Conference Proceeding
Language:English
Subjects:
Accuracy
Alzheimer's disease
Brain complexity
Brain modeling
classification
Coherence
Electroencephalography
Entropy
machine learning
magnitude squared coherence
multiscale fuzzy entropy
Reproducibility of results
support vector machine
Support vector machines
synchronization
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Alzheimer's disease (AD) is a progressive neurode-generative condition that impacts cognitive functions and the overall quality of life of millions of individuals worldwide. Early and accurate diagnosis of AD is crucial for providing appropriate treatment. In this paper, an algorithm that uses a linear support vector machine (SVM) to classify electroencephalographic (EEG) signals of healthy subjects and patients with AD is proposed. The features used for the classification are the multiscale fuzzy entropy (MFE) as a measure of brain complexity and the magnitude-square coherence (MsCoh) as a synchronization measure between different brain regions, considering four frequency bands: delta, theta, alpha, and beta. The classification results for AD and HS were obtained through the leave one subject out cross-validation (LOOVC), and the model's performance was evaluated by using classification accuracy, sensitivity, and specificity. Specifically, a mean classification accuracy of 93.6% with a repeatability of 1.6% and a reproducibility of 6.5% was obtained. Furthermore, Shapley values were computed to explain the output of the classification model. To validate the proposed method, data from a publicly available dataset was utilized. This preliminary study demonstrates that SVM is a powerful tool for diagnosing AD by using the complementarity of complexity and synchronization features of EEG signals.
ISSN:2642-2077
DOI:10.1109/I2MTC60896.2024.10560664