Detection of cognitive load using EEG signal and lifting wavelet transform with specific lead selection

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Mental arithmetic tasks involve complex cognitive processes.
• Analyzing brain responses, particularly electroencephalographic (EEG) signals, is crucial for diagnosing diseases and understanding stress.
• Reducing the number of EEG leads can enhance cost-effectiveness and reduce complexity in cognitive load detection.

Purpose of the Study:

• To develop an effective approach for recognizing mental arithmetic load using minimal EEG leads.
• To evaluate the performance of a cost-effective, reduced-complexity technique for cognitive load detection.
• To assess the feasibility of using only two frontal EEG leads (Fp1 and Fp2) for accurate classification.

Main Methods:

• Utilized electroencephalographic (EEG) signals from two frontal leads (Fp1 and Fp2).
• Applied lifting wavelet processing to divide EEG signals into 12 frequency band segments.
• Extracted fuzzy entropy features and employed the lowest redundancy maximum relevance technique for feature selection.
• Compared three supervised machine learning models: K-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Random Forest Algorithm (RFA).

Main Results:

• The Support Vector Machine (SVM) classifier achieved the highest accuracy of 96.63% when using 19 EEG leads.
• With only two frontal leads (Fp1 and Fp2), the SVM classifier still provided the highest accuracy at 95.34%.
• The proposed technique demonstrated preferable classification accuracy with a reduced number of EEG leads.

Conclusions:

• The developed technique effectively recognizes mental arithmetic load using a minimal number of EEG leads.
• The method is suitable for designing cost-effective wearable devices for cognitive load detection.
• Reduced lead EEG analysis offers a practical approach for real-time cognitive monitoring.