Comparison of Fractal Dimension and Wavelet Transform Methods in Classification of Stress State from EEG Signals

Abstract

Stress is a significant issue in everyday life that affects both physical and mental health. There are different approaches to stress classification. This research examines the implementation of the fractal dimension (FD) method as one of the features for stress state classification using brain signals. Consequently, the comparison between FD and wavelet transform has been conducted using electroencephalogram (EEG) signals recorded during the Stroop Colour Word Test (SCWT). The comparison results show that the FD is better in the classification of the stress state. The highest F1 score has been obtained using FD with quadratic support vector machine (SVM) in average 83.03% for the comparison between baseline session and different stress states. Besides, FD with medium Gaussian SVM has the highest F1 score, on average 83.36%, for comparison between various stress states.