Prevention of spinal cord injury with time-frequency analysis of evoked potentials: an experimental study

Abstract

Results-The TFDs of EPs were found to concentrate in a certain location under normal conditions. When injury occurred, the energy decreased in peak power, and there was a greater dispersion of energy across the time-frequency range. Strong relations were found between latency and peak time, and amplitude and peak power. However, the change in peak power after injury was significantly larger than the corresponding change in amplitude (p<0.001 by ANOVA). Conclusions-It was found that TFA of EPs provided an earlier and more sensitive indication of injury than time domain monitoring alone. It is suggested that TFA of EP signals should therefore be useful in preventing spinal cord injury during surgery. (J Neurol Neurosurg Psychiatry 2001;71:732-740) Keywords: time frequency analysis; evoked potential; spinal cord injury Spinal surgery is a common form of treatment for various spinal deficits, but also entails the risk of spinal cord damage, which may lead to sensory or motor problems, and even paralysis. To reduce this risk, various evoked potential techniques have been used to monitor the integrity of spinal cord function during surgery. However, false outcomes from monitoring are still a concern to the monitoring and surgical teams. 1 Current intraoperative monitoring techniques measure only the latency and amplitude of the evoked potential (EP) 1 2 despite the fact that EP signals are usually polyphasic waveforms that reflect diVerent activation and conduction velocities within the spinal cord. 3 Present measurements cannot therefore represent the precise characteristics of EP signals. The power spectrum of EP signals can indicate the proportion of diVerent frequency components in the signal, and may be more representative of the physical nature of the signal than the features seen in the time spectrum alone. As such, combining time and frequency analysis into time-frequency analysis (TFA) could be a useful tool in spinal cord monitoring. 4 The main advantage of this is the integrity of information from the whole of the EP waveform, rather than arbitrarily selected features.Application of TFA to EP signals to monitor the integrity of spinal cord function during spinal surgery has not yet been reported. In the present study, the changes of various EP waveforms in time-frequency space were studied after mechanical insult to the spinal cord. The purpose of this was to assess the applicability of TFA methods to EP signals, and provide a basis for the clinical use of TFA in spinal EP monitoring. Materials and methods EXPERIMENTAL PROCEDURETwenty mature rats weighing between 260 and 280 g were used. All the surgical procedures were performed under intravenous pentobarbital (0.05 mg/g) anaesthesia augmented by local 1% xylocaine infiltration. Additional pentobarbital was given at intervals and in amounts established in non-curarised rats to assure adequate anaesthesia.Many investigators use physically transected spinal cords for studying spinal cord regeneration, but this is rarely encountered in injury to the human ...

Comparison of time-frequency distribution techniques for analysis of spinal somatosensory evoked potential

“…Another earlier study specifically examined the optimum trade-off between time and frequency resolution by using STFT on somatosensory, visual, and brainstem auditory evoked potentials (DE WEERD and KAP, 1981). The application of TFA to SSEP signals can provide a reliable indication of spinal cord function (Hu et al, 2000). However, different TFA algorithms have their own fitting conditions, and algorithms should be adapted to match the definitive characteristics of the signal.…”

“…The SSEP signal was interpreted in practice by displaying a two-dimensional TFA distribution, and three parameters previously found to be indicative of spinal cord injury were recorded: peak time, peak frequency and peak power (BRAUN et al, 1997;HU et al, 2000). The accuracy of spinal cord monitoring therefore depends on the clear interpretation of the time-frequency distribution and accurate energy representation.…”

“…As such, the resolutions of the peaks in the time-frequency distribution were used as the criteria for comparison between methods. Pilot studies have shown that the lower frequency peak in the time-frequency plot of SSEP is a more reliable indicator of injury-related TFA change than the higher frequency peaks (Hu et al, 2000) and therefore this peak was used as the basis for comparison between the methods. To choose a practical TFA method for intraoperative SSEP monitoring, all the signals were analysed with the various algorithms, including different window functions and lengths.…”

“…To obtain information on the frequency domain as well as time domain, time-frequency analysis (TFA) methods must be used. TFA ofEP waveforms has demonstrated that the peak energy in the TFD may be of particular diagnostic value in assessing neurological injury (Hu et al, 2000). The energy of the EP waveforms is normally found to be concentrated in a certain time-frequency space, and this can be used to interpret the character of the EP signal in greater detail and more reliably than in the time domain alone (BRAUN et al, 1997;HU et al, 2000).…”

“…TFA ofEP waveforms has demonstrated that the peak energy in the TFD may be of particular diagnostic value in assessing neurological injury (Hu et al, 2000). The energy of the EP waveforms is normally found to be concentrated in a certain time-frequency space, and this can be used to interpret the character of the EP signal in greater detail and more reliably than in the time domain alone (BRAUN et al, 1997;HU et al, 2000). injury is known to cause localised changes in the morphology of the SSEP waveform, and therefore in the timefrequency spectrum of the EP waveforms (BRAUN et al, 1997;HU et al, 2000).…”