Weighted ordinal connection based functional network classification for schizophrenia disease detection using EEG...

Area of Science:

• Neuroscience
• Computational Neuroscience
• Biomedical Engineering

Background:

• Brain connectivity networks (BCN) are vital for understanding brain function.
• Predictability of BCNs depends heavily on the chosen connectivity measure.
• Selecting appropriate functional connectivity metrics is crucial for clinical and cognitive neuroscience.

Purpose of the Study:

• To identify suitable connectivity measures for electroencephalogram (EEG)-based BCN.
• To propose an efficient network identifier for distinguishing brain states.
• To evaluate the performance of different classifiers on EEG data from schizophrenia patients.

Main Methods:

• Constructed weighted BCNs (WBCNs) using correlation coefficient (r), coherence (COH), phase-locking value (PLV), and mutual information (MI) from EEG signals.
• Applied weighted ordinal connections for feature extraction.
• Utilized k-nearest neighbours (KNN), support vector machine (SVM), random forest (RF), and 1D convolutional neural network (CNN1D) for classification.

Main Results:

• Achieved 90% classification accuracy using a 1D CNN classifier with WBCN based on the coherence connectivity measure.
• Demonstrated the effectiveness of coherence as a superior connectivity metric for EEG-based BCN.
• Provided a structural analysis of the BCN.

Conclusions:

• Coherence is an effective connectivity measure for building predictive BCNs from EEG data.
• The proposed method, utilizing coherence and 1D CNN, accurately classifies brain states.
• This approach holds promise for advancing diagnostic tools in neuroscience.