Measuring latency variations in evoked potential components using a simple auto-correlation technique

Interpretation of averaged evoked potentials is difficult when the time relationship between stimulus and response is not constant. Later components are more prone to latency jitter, making them insufficiently reliable for routine clinical use even though they could contribute to greater understanding of the functioning of polysynaptic components of the afferent nervous system. This study is aimed at providing a simple but effective method of identifying and quantifying latency jitter in averaged evoked potentials. Autocorrelation techniques were applied within defined time windows on simulated jittered signals embedded within the noise component of recorded evoked potentials and on real examples of somatosensory evoked potentials. We demonstrated that the technique accurately identifies the distribution and maximum levels of jitter of the simulated components and clearly identifies the jitter properties of real evoked potential recording components. This method is designed to complement the conventional analytical methods used in neurophysiological practice to provide valuable additional information about the distribution of latency jitter within an averaged evoked potential. It will be useful for the assessment of the reliability of averaged components and will aid the interpretation of longer-latency, polysynaptic components such as those found in nociceptive evoked potentials.