Detection of pain from nociceptive laser-evoked potentials using single-trial analysis and pattern recognition

Pain is an unpleasant multidimensional experience, which could be largely influenced by various peripheral and cognitive factors. Therefore, the pain experience and the related brain responses exhibit high variability from time to time and from condition to condition. The availability of an objective assessment of pain perception would be of great importance for both basic and clinical applications. In the present study, we combined single-trial analysis and pattern recognition techniques to differentiate nociceptive laser-evoked brain responses (LEPs) and resting electroencephalographical recordings (EEG). We found that quadratic classifier significantly outperformed linear classifier when separating LEP trials from resting EEG trials. Across subjects, the error rates of quadratic classifier, when it was tested on all trials (I1+I2), trials with low ratings (I1), and trials with high rating (I2), are respectively 17.5±3.5%, 20.6±4.3%, and 9.1±4.9%.