Diverse frequency band-based convolutional neural networks for tonic cold pain assessment using EEG

The purpose of this study is to present a novel classification framework, called diverse frequency band-based Convolutional Neural Networks (DFB-based ConvNets), which can objectively identify tonic cold pain states. To achieve this goal, scalp EEG data were recorded from 32 subjects under cold stimuli conditions. The proposed DFB-based ConvNets model is capable of classifying three classes of tonic pain: No pain, Moderate Pain, and Severe Pain. Firstly, the proposed method utilizes diverse frequency band-based inputs to learn temporal representations from different frequency bands of Electroencephalogram (EEG) which are expected to have more discriminative power. Then the derived features are concatenated to form a feature vector, which is fed into a fully-connected network for performing the classification task. Experimental results demonstrate that the proposed method successfully discriminates the tonic cold pain states. To show the superiority of the DFB-based ConvNets classifier, we compare our results with the state-of-the-art classifiers and show it has a competitive classification accuracy (97.37%). Moreover, these promising results may pave the way to use DFB-based ConvNets in clinical pain research.