Classification of stress and non-stress condition using functional near-infrared spectroscopy

In these days, human health is most affected by the chronic stress, which is a well-known cause of several diseases such as heart disease, obesity, and stroke. The brain imaging modalities can be used as a viable option for the indirect assessment of the stress. In this study, we have used functional near-infrared spectroscopy (fNIRS) to discriminate between the hemodynamic responses of two different conditions (i.e., stress and non-stress). The Montreal Imaging Stress Task (MIST) is used to induce stress in the participants. fNIRS signals related to the stress and non-stress conditions are measured from the prefrontal cortex. The active channels are selected by determining the t-value between the measured fNIRS data and the desired hemodynamic response function. Linear discriminant analysis (LDA) is used to classify the stress and non-stress condition based on the combinations of various features (mean, peak, slope, kurtosis, and skewness values). The results show that the combination of slope and kurtosis yielded the highest classification accuracy of 70% between the stress and non-stress condition. Our preliminary results indicated that fNIRS has an ability to measure and classify stress and non-stress condition of humans.