EOG Signal Classification with Wavelet and Supervised Learning Algorithms KNN, SVM and DT.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Machine Learning

Background:

• Electrooculography (EOG) records electrical potential changes from eye movements.
• Eye movements generate physiological signals crucial for understanding visual attention and neurological conditions.
• Accurate classification of EOG signals is essential for developing assistive technologies and diagnostic tools.

Purpose of the Study:

• To classify five distinct human eye movements (left, right, down, up, blink) using EOG signals.
• To evaluate the performance of different supervised learning algorithms for EOG signal classification.
• To identify the most effective machine learning model for accurate EOG-based eye movement detection.

Main Methods:

• EOG signals were analyzed using Wavelet Transform for frequency domain characterization (0.5-50 Hz bandwidth).
• Supervised learning algorithms, including K-Nearest Neighbor (KNN), Support Vector Machine (SVM), and Decision Tree (DT), were implemented for classification.
• Classification accuracy was assessed using metrics such as the Jaccard index, confusion matrix, and Receiver Operating Characteristic (ROC) curve.

Main Results:

• The Support Vector Machine (SVM) classifier achieved the highest accuracy of 76.9%.
• K-Nearest Neighbor (KNN) and Decision Tree (DT) classifiers showed lower accuracies at 69.4% and 60.5%, respectively.
• The Jaccard index confirmed SVM as the superior classifier for this EOG signal classification task.

Conclusions:

• Support Vector Machine (SVM) is the most effective algorithm for classifying electrooculography (EOG) signals corresponding to five eye movements.
• Wavelet Transform provides valuable frequency domain features for characterizing EOG signals.
• Accurate EOG signal classification holds promise for advanced human-computer interaction and clinical diagnostics.