Evaluation of an automated analysis for pain-related evoked potentials

Abstract

This paper presents initial steps towards an automated analysis for pain-related evoked potentials (PREP) to achieve a higher objectivity and non-biased examination as well as a reduction in the time expended during clinical daily routines. While manually examining, each epoch of an ensemble of stimulus-locked EEG signals, elicited by electrical stimulation of predominantly intra-epidermal small nerve fibers and recorded over the central electrode (Cz), is inspected for artifacts before calculating the PREP by averaging the artifact-free epochs. Afterwards, specific peak-latencies (like the P0-, N1 and P1-latency) are identified as certain extrema in the PREP’s waveform. The proposed automated analysis uses Pearson’s correlation and low-pass differentiation to perform these tasks. To evaluate the automated analysis’ accuracy its results of 232 datasets were compared to the results of the manually performed examination. Results of the automated artifact rejection were comparable to the manual examination. Detection of peak-latencies was more heterogeneous, indicating some sensitivity of the detected events upon the criteria used during data examination.