Fig 1.
Generation of template signals used for waveform similarity analysis.
Sural nerve (A) and tibial muscle EMG (B) template signals were generated from the intact side (left) of rats. The mean for each template was derived after individual responses were temporally aligned to the primary positive peak (arrows). See text for further details.
Fig 2.
Example of event detection, latency and response magnitude of sural CNAPs derived by WSA.
Row A illustrates three N10avRs (repeats are colour coded) to a 0.7 mA stimulus of an intact (A1) and regenerating (A2) nerve from the same animal, with the SN-template (see Fig 1A) superimposed (black). The cross-correlation was calculated for each instant between the SN-template and the N10avR, by sliding the SN-template trace across each N10avR. Row B illustrates the mean (black) and one standard deviation (red) of the cross-correlations calculated between SN-template and the three repeated N10avRs for an intact (B1) and a regenerating (B2) nerve of the same animal. An event was considered detected when the peaks of the mean of 3 cross-correlations was ≥ a detection threshold level (see Fig 3 for calculation of the detection threshold level). The latency for each event was given by the time of each respective mean cross-correlation peak (arrows row B) plus the 0.35 ms time lag between the onset and peak of SN-template (grey bars indicate latency correction, row A). The magnitude of each event was quantified as the peak of the mean cross-correlation for each event detected (arrows, B1 and B2), and is proportional to the energy of the event measured in arbitrary units (amplitude correction not shown). Total magnitude of each N10avR was calculated as the sum of the magnitudes of all events in N10avR (sum of all peaks indicated by arrows in B1 and B2).
Fig 3.
Effect of WSA event detection threshold levels on the number of identified events.
Empirically derived functions were used to determine an ideal detection threshold. A demonstrates the mean number of events detected by WSA from N10avRs evoked from injured sciatic nerves by a 0.7 mA stimulus, as a function of the detection threshold level. The detection threshold level indicates the multiple of standard deviations (refer to red line in Fig 2B2 showing one standard deviation) that the mean (black line, Fig 2B2) of 3 repeated N10avRs had to be greater than, in order to detect an event using WSA. The fuzzy line represents the mean and SEM (6.5 ± 0.6) of the number of events detected by TEA for the same responses. Throughout the study, 2.0× was chosen as the standard detection threshold level for N10avRs, as this level detects a number of events not significantly different to that detected by TEA. B demonstrates the mean number of events detected from M5avRs evoked from injured sciatic nerves by a paired stimulus as a function of background noise level. The paired stimulus was 0.7 mA with an inter-pulse interval of 1 ms. Background noise was calculated from the mean absolute amplitude of the last 17 ms of the sweep, a region where no signal was present. The detection threshold level indicates the threshold based on a multiple of background noise. The grey line indicates the ideal level of event detection, i.e. where the number of events = 1. For in vivo experiments, 12× the background noise level was chosen as the standard detection threshold level for M5avRs.
Fig 4.
Pre-processing and quantification of CMAP responses by WSA and TEA.
Shown are examples of paired conditioning (black arrow, row A) and test (white arrow, row A) stimuli presented to an intact sciatic nerve, with inter-pulse intervals of 1 ms (column 1) and 10 ms delays (column 2) of the same animal. First, an average of 2 repeated M5avRs were evoked by a single conditioning pulse stimulus (0.7 mA 0.2 ms) at 5.0 ms after the onset of the recording sweep (black arrow, row A) in the absence of a following test pulse. This average trace (dotted traces, row A) was subsequently subtracted from each M5avR evoked by the paired stimuli (grey traces, row A) to reveal the response to the test pulse in the absence of the conditioning pulse (row B), and subsequently quantified by TEA (peak-to-peak, arrows, row B) and WSA methods (rows C and D). The template comparing wave, M-template (black trace, row C) was slid across the pre-processed M5avR (dotted trace, row C) to derive the cross-correlation function between the M-template and the pre-processed M5avR signals at each time instant (row D, positive cross-correlations black, negative cross-correlations dotted). Positive peak values of the cross-correlation sequence represent arbitrary units (a.u.) of the event magnitude (amplitude correction not shown), and the latency is measured as the time at the cross-correlation peak plus the 3.5 ms time lag (latency correction, indicated by double arrows, row D) between the onset and peak of M-template (double arrows, rows C).
Fig 5.
Simulation experiment of WSA performance under conditions of template distortion.
Simulated signals generated by adding three multiples (0.5×, 1.0× and 2.0×) of a signal of interest (A, true signal) to noise (B), were used to derive a signal for analysis (C, total). WSA was applied using three templates; the correct one identical to 1.0× the true signal (C, blue), a template with double (C, red) and half (C, green) the frequency content of the correct template. Latency-corrected WSA outputs calculated from the cross-correlation sequence (xycorr) are shown (D) for each template (WSA1, blue, output with correct template; WSA2, red, template with double the correct frequency; WSA3, green, template with half the correct frequency). Overlayed to WSA outputs is the true signal, with the positive and negative peaks indicated (black dots) that reports the true peak-to-peak measurements for comparing the performance of WSA and TEA. Peak detection threshold at >4× the standard deviation of the first 400 ms of each respective WSA output was used to automate peak detection (D, blue, red and green dots indicate peaks for WSA1, WSA2 and WSA3 respectively). The latency (E) and Magnitude (F) errors are shown for WSA performed by each template (following latency and amplitude correction respectively) and TEA for the 6 samples of each signal. The black line indicates zero error; data expressed as mean ± SEM.
Fig 6.
Latency profiles derived by WSA and TEA.
The latency to the peak of the first event of N10avRs (A1, intact side; A2, regenerating side) and the latency to the peak of M5avRs (B1, intact side; B2, regenerating side) were measured using the temporal location of best fit between the event and its respective template signals (WSA), and compared with the latency of the peak of each respective event (TEA). There was no significant difference in the latencies derived from WSA and TEA. Inter-pulse interval = time interval between the conditioning and testing pulses (refer to Fig 4). Abbreviations: WSA, waveform similarity analysis; TEA, trained eye analysis.
Fig 7.
Magnitude profiles derived from WSA and TEA.
N10avR total magnitudes (TMags) were calculated by quantifying and summing magnitudes of all events for each response of intact (A1) and regenerating (A2) sural nerves using WSA (black) and TEA (grey). TMags derived from WSA were amplitude-corrected (dashed) by multiplying the TMagWSA by a constant (indicated) derived from the ratio of [mean TMagTEA]/[mean TMagWSA] (see Materials and Methods) evoked from the strongest stimulus pulse (1.4 mA). Magnitude profiles were also derived from M5avR TMags for intact (B1) and regenerating (B2) nerves. Similarly, TMagWSA of M5avRs were amplitude-corrected (dashed) by multiplying by a constant (indicated) based on the ratio of mean total magnitudes with the greatest inter-pulse interval (10 ms).