Epileptic spasm recognition: EEG classification using time-frequency features and machine learning.

Area of Science:

• Neurology
• Biomedical Engineering
• Data Science

Background:

• Epileptic spasm (ES) presents diagnostic challenges, particularly in pediatric populations.
• Current EEG-based seizure detection methods struggle with the diverse patterns of ES.
• Accurate and automated classification of ES is crucial for timely intervention.

Purpose of the Study:

• To develop and evaluate a machine learning approach for classifying epileptic spasm (ES) EEG signals.
• To investigate the effectiveness of time-frequency domain features for ES classification.
• To compare the performance of Random Forest, KNN, and SVM models in ES detection.

Main Methods:

• Clinically collected EEG data from patients with epileptic spasms were analyzed.
• A set of 54 time-frequency domain features were extracted from the EEG signals.
• Machine learning models including Random Forest (RF), K-Nearest Neighbors (KNN), and Support Vector Machine (SVM) were trained and tested.

Main Results:

• The Random Forest model achieved the highest classification accuracy of 81.18% with a reduced feature set.
• K-Nearest Neighbors (KNN) demonstrated improved performance with an increased number of features.
• The study successfully classified ES EEG patterns using the proposed feature extraction and machine learning techniques.

Conclusions:

• Combining time-frequency features with machine learning models shows significant potential for accurate epileptic spasm classification.
• The developed approach offers a promising tool for automated monitoring and diagnosis of ES.
• Further research is recommended to enhance feature extraction and model robustness for clinical implementation.