Differentiation of pain levels by deploying various electroencephalogram synchronization features and a dynamic ensemble selection mechanism

Objective: The aim of this study was to measure pain intensity in an objective manner by analyzing electroencephalogram (EEG) signals. Although this problem has attracted the attention of researchers, increasing the resolution of this measurement by increasing the number of pain states significantly decreases the accuracy of pain level classification. Approach: To overcome this drawback, we adopt state-of-the-art synchronization schemes to measure the linear, nonlinear and generalized synchronization between different EEG channels. Thirty-two subjects executed the cold pressor task and experienced five defined levels of pain while their EEGs were recorded. Due to the large number of synchronization features from 34 channels, the most discriminative features were selected using the greedy overall relevancy method. The selected features were applied to a dynamic ensemble selection system. Main results: Our experiment provides 85.6% accuracy over the five classes, which significantly improves upon the results of past research. Moreover, we observed that the selected features belong to the channels placed over the ridge of the cortex, the area responsible for processing somatic sensation arising from nociceptive temperature. As expected, we noted that continuation of the painful stimulus for minutes engaged regions beyond the sensorimotor cortex (e.g. the prefrontal cortex). Significance: We conclude that the amount of synchronization between scalp EEG channels is an informative tool in revealing the pain sensation.