EEG-Based Emotion Assessment using Detrended Flunctuation Analysis (DFA)

Abstract

Stroke patients suffer from emotional and behavioral changes. The emotion assessment of stroke patients is helpful to carry out appropriate treatment. Emotion assessment through Electroencephalogram (EEG) is reliable and can be applied to stroke patients. Fractal analysis using Detrended Fluctuation Analysis (DFA) is applied to detect the temporal correlation and the simplicity of EEG signals. Emotion contained-EEG signals of two groups of stroke patients, with left brain damage (LBD) and right brain damage (RBD), and a group of normal control (NC) were assessed using DFA in alpha, beta and gamma frequency bands. The EEG signals of the three groups show different degrees of temporal anti-correlation. Moreover, alpha and beta bands which exhibit larger brain oscillation have better performance in emotion classification than gamma band. The overall performance of DFA has achieved 92.00% classification accuracy in LBD, and 91.75% in RBD. Thus, DFA is useful in emotion assessment of stroke patients.