Facial expressions in correlation with FNIRS during stroop task experiment

Abstract

This thesis aims to anticipate the correlation between the hemodynamic responses of frontal cortex and facial expressions during neurocognitive training task, Stroop Task experiment to understand emotion elicited from particular tasks including stressful events. In the experiment, 20 participants between the ages of 18 - 22 volunteer to be involved in the study of Stroop Task experiment whereas the database of fNIRS signal and facial expressions are captured continuously. The workstream requires preprocessing of both fNIRS data and RGB-D video, extracting features from fNIRS signal and detection of facial expressions. Both features are then classified using k-Nearest Neighbor (kNN) algorithm. The General Linear Model of hemodynamic response is established results are shown by the estimation of GLM parameters and activation mapping on the cerebral region of interest. The correlation of fNIRS signal and facial expressions using Person’s correlation coefficient with significance of approximately 55% which can further be inspected in results. This correlation is found to be between moderate to strong relationship with the p value in the range of .001 to .071. The results is considered to be significant in research domains like psychology and many other applications of Brain Computer Interface (BCI). Keywords Near Infrared Spectroscopy (NIRS), Facial Expressions, Stroop Task Experiment, Emotion Detection, Correlation.