Application of a novel automatic duration method measurement based on the wavelet transform on pathological motor unit action potentials.

Objective: To evaluate a recently published automatic duration method based on the wavelet transform applied on normal and pathological motor unit action potentials (MUAPs).

Methods: We analyzed 313 EMG recordings from normal and pathological muscles during slight contractions. After the extraction procedure, 339 potentials were accepted for analysis: 68 from normal muscles, 124 from myopathic muscles, 20 from chronic neurogenic muscles, 83 from subacute neurogenic muscles and also 44 fibrillation potentials, as an example of very low duration muscular potentials. A "gold standard" of the duration positions (GSP) was obtained for each MUAP from the manual measurements of two senior electromyographists. The results of the novel method were compared to five well-known conventional automatic methods (CAMs). To compare the six methods, the differences between the automatic marker positions and the GSP for the start and end markers were calculated. Then, for the different groups of normal and pathological MUAPs, we applied: a one-factor ANOVA to compare their relative mean differences, the estimated mean square error (EMSE) and a Chi-square test about the rate of automatic marker placements with differences to the GSP greater than 5 ms, taken as gross errors.

Results: The mean and the standard deviation of the differences, the EMSE and the gross errors for the novel method were smaller than those observed with the CAMs in the five different MUAP groups and significantly different in most of the cases.

Conclusions: The novel automatic duration method is more accurate than other available algorithms in normal and pathological MUAPs.

Significance: Accurate MUAP duration automatic measurement is an important issue in daily clinical practice.