EEG-Based Detection of Mild Cognitive Impairment Using DWT-Based Features and Optimization Methods.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Electroencephalography (EEG) is increasingly explored for identifying neurological disorders.
• Accurate detection of mild cognitive impairment (MCI) is crucial for timely intervention.
• Traditional EEG analysis often involves numerous channels, posing challenges for clinical application.

Purpose of the Study:

• To develop and validate discrete wavelet transform (DWT)-based biomarkers for detecting MCI from EEG signals.
• To investigate the impact of reducing EEG channel count on MCI classification accuracy using multi-objective optimization.
• To assess the feasibility of a reduced-channel EEG system for clinical MCI diagnosis.

Main Methods:

• EEG signals from 29 MCI patients and 32 healthy subjects were analyzed.
• Discrete Wavelet Transform (DWT) was applied to decompose EEG signals and extract non-linear features.
• Machine learning classifiers were trained on extracted features.
• Multi-objective optimization algorithms (NSGA-II, NSGA-III, PSO) were employed for optimal channel selection.

Main Results:

• DWT-based features achieved high classification accuracy with full EEG channel usage (99.84%).
• Optimized selection of fewer channels, particularly using NSGA-II, NSGA-III, or PSO, enhanced accuracy to 99.97% with four channels.
• A five-channel selection using NSGA-II achieved a perfect classification accuracy of 100%.

Conclusions:

• DWT-based feature extraction offers a promising approach for MCI detection using EEG.
• Strategic reduction and selection of EEG channels can improve, not hinder, classification accuracy.
• A minimal set of optimized EEG channels holds significant potential for cost-effective and efficient clinical diagnosis of MCI.