Decoding of Subjective Pain-Sensitivity by Brain Signal Analysis Using a General Type-2 Fuzzy Classifier

The prime mechanism governing the variability of pain perception across different subjects is still unexplored. This paper intends to develop a novel methodology to investigate this phenomenon using EEG signal analysis system. First, the EEG signals are procured from the scalp of subjects who are presented with three types of touch stimuli: heat, bristles and pinch with varying intensity levels. The raw brain signals acquired are analyzed using eLORETA software that confirms the involvement of primary somatosensory cortex and dorsal region of anterior cingulate cortex for this cognitive activity. Additionally, frequency analysis undertaken infers the participation of delta, alpha and theta bands for the said task. The signals are then transferred to a feature extraction module where a dual feature extraction strategy has been employed using Power Spectral Density (PSD) and Discrete Wavelet Transform (DWT) to enhance the diversity of the feature set. The abstracted features are further evaluated using Principal Component Analysis (PCA) to retain the most important or optimal features. The reduced feature set is transferred to a novel General Type-2 fuzzy classifier that is able to precisely classify the distinct class labels and also outperforms its conventional counterparts. Hence, this method can help to assess the variability of pain perception amongst individuals whose communication modality is crippled due to scenarios pertaining to neurological disorders, anaesthetic treatments and the like. Moreover, the present scheme can be utilized as a neuronal marker to distinguish individuals suffering from extreme sensitivity towards pain from the healthy ones.