Measuring Latency Variations in Evoked Potential Components Using a Simple Autocorrelation Technique

Figure 1

Flow chart of the autocorrelation method. single response recordings of the evoked potential are used. For each possible, dissimilar pair of recordings a and b (of which there are combinations), an autocorrelation is performed within a specified time window ( to ) by calculating the cross-correlation between the recordings (, where identifies the unique pair of recordings). Recording b is then shifted along the time axis (relative to recording a) by a small amount (usually the dwell time of the recording), and the correlation is again calculated. This is repeated until recording b is shifted by the length of the time window. The time shift that produced the largest is then recorded, where represents the latency shift between that pair of recordings. Another pair of recordings is then selected, and the process is repeated until all dissimilar pairs of recordings have been autocorrelated. The histogram of the s for all possible pairs shows the distribution of the latency jitter in the set of responses.