Time-frequency Feature of Intraoperative Somatosensory Evoked Potential Signals

The purpose of this paper is to investigate the time-frequency analysis of somatosensory evoked potentials (SEP) and its potential application to intraoperative spinal cord monitoring. In this study, SEP signals were recorded from 97 adolescent patients undergoing surgical correction of idiopathic scoliosis. The time-frequency distributions of SEP were observed during surgery by analysing averaged SEP signals with short time Fourier transform (STFT). Main peak in time-frequency interpretation of SEP was measured in peak time, peak frequency and peak power. The changes of these parameters were compared with the changes of latency and amplitude. The results showed that changes in peak times and peak powers were found to correlate to the changes of latency and amplitude, respectively. However, the peak time showed more variability than the latency (p<0.01), while the peak power showed significantly lower variability than the amplitude (p<0.01). The peak frequency of SEP seems to be unchanged during surgery. Time-frequency analysis of SEP waveform gives stable and easily identifiable characteristics, and also shows higher specificity than amplitude and latency. Applying time-frequency analysis to SEP may improve the reliability of intraoperative spinal cord monitoring.