Fig. 2
From: Detection of spinal action potentials with subdural electrodes in freely moving rodents
Schematic of the propagation algorithm. (a) A representative spike is present on all electrodes, with the middle electrodes (E5, E6 and E7) highlighted to further illustrate the algorithm in panels b-d. (b) Cross-correlation between the waveforms on E5 and E6 (top row), and E6 and E7 (bottom row). Delays of 0, 1 and 2 data points are shown. For the correlation between E6 and E5, a delay of 1 aligns these spikes the best, whereas for E7 relative to E6, a delay of 2 is best. Cross-correlation is carried out between all pairs of electrodes to create a matrix, with the two examples shown in purple and orange. (c) Each column is shifted to align the top channel as the reference, creating a row of zeros. The mean is calculated for each row to determine an average delay between each channel and the reference channel, resulting in a single delay for each electrode. (d) A linear model is fitted between the average time delays calculated in c) and the known distances between the electrodes. The gradient of this linear model represents the velocity of the given spike.
