Combination of Empirical Mode Decomposition Components of HRV Signals for Discriminating Emotional States

Document Type: Original Paper

Authors

Computational Neuroscience Laboratory, Department of Biomedical Engineering, Faculty of Electrical Engineering, Sahand University of Technology, Tabriz, Iran

Abstract

Introduction
Automatic human emotion recognition is one of the most interesting topics in the field of affective computing. However, development of a reliable approach with a reasonable recognition rate is a challenging task. The main objective of the present study was to propose a robust method for discrimination of emotional responses thorough examination of heart rate variability (HRV). In the present study, considering the non-stationary and non-linear characteristics of HRV, empirical mode decomposition technique was utilized as a feature extraction approach.
Materials and Methods
In order to induce the emotional states, images indicating four emotional states, i.e., happiness, peacefulness, sadness, and fearfulness were presented. Simultaneously, HRV was recorded in 47 college students. The signals were decomposed into some intrinsic mode functions (IMFs). For each IMF and different IMF combinations, 17 standard and non-linear parameters were extracted. Wilcoxon test was conducted to assess the difference between IMF parameters in different emotional states. Afterwards, a probabilistic neural network was used to classify the features into emotional classes.
Results
Based on the findings, maximum classification rates were achieved when all IMFs were fed into the classifier. Under such circumstances, the proposed algorithm could discriminate the affective states with sensitivity, specificity, and correct classification rate of 99.01%, 100%, and 99.09%, respectively. In contrast, the lowest discrimination rates were attained by IMF1 frequency and its combinations.
Conclusion
The high performance of the present approach indicated that the proposed method is applicable for automatic emotion recognition.

Keywords

Main Subjects