Lattice 123 pattern for automated Alzheimer's detection using EEG signal.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Alzheimer's disease (AD) diagnosis relies on sensitive and accurate methods.
• Electroencephalogram (EEG) signals offer a non-invasive window into brain activity.
• Current feature extraction techniques for EEG-based AD detection require enhancement.

Purpose of the Study:

• To develop an innovative feature engineering framework for automated Alzheimer's disease identification using EEG signals.
• To introduce the Lattice123 pattern for enhanced textural feature extraction from EEG data.
• To achieve high classification accuracy for AD detection through advanced signal processing and machine learning.

Main Methods:

• A novel Lattice123 pattern was engineered using probabilistic functions inspired by Shannon information entropy.
• Directed graphs with distance-based kernels and multiple kernel functions were utilized for feature vector generation.
• Multilevel discrete wavelet transform (MDWT) was employed for low-level wavelet subband generation.
• Iterative neighborhood component analysis and voting algorithms were used for feature selection and classification.

Main Results:

• The proposed framework achieved a classification accuracy exceeding 98%.
• A geometric mean of over 96% was obtained, indicating robust performance.
• The Lattice123 pattern effectively extracted subtle changes in EEG signals, crucial for AD detection.

Conclusions:

• The developed Lattice123 pattern and feature extraction framework provide an accurate method for Alzheimer's disease detection from EEG signals.
• The approach demonstrates potential for early and reliable AD diagnosis.
• The prototype is prepared for validation with extensive and diverse datasets.