Calculating Turns/Amplitude peak ratio of EMG interference pattern by using logistic curve fitting

Abstract

The aim of the study was to test whether logistic curve fitting (LCF) of Turns = f(Amplitude) plots of single muscle contractions can provide a reliable alternative method for peak-ratio calculation. EMG signals from 74 biceps and 62 triceps contractions were analyzed by applying LCF to Turns = f(Amplitude) plots. Peak-ratio (peak-ratio2) could then be calculated as the point of the fitted line with the highest Turn/Amplitude value. LCF yielded R2 values > 0.95 in the vast majority of contractions studied (68/74 biceps and 53/62 triceps). Peak-ratio2 values had a very strong linear relationship with the corresponding values calculated by the traditional method (peak-ratio1) in both normal and neurogenic conditions. Furthermore, ROC curve analysis showed that peak-ratio1 and peak-ratio2 had similar AUC values. Based on the LCF equation, peak-ratio = T2*(p − 1)/A0*p*(p − 1)1/p. Therefore, peak-ratio is proportional to the maximum number of turns (T2), positively correlated to the rate of turns’ increment at the midpoint of the curve (p) and negatively correlated to the mean amplitude at the midpoint of the curve (A0). A0 is the variable that best discriminates between normal and neurogenic conditions. We provide an alternative method for peak-ratio calculation and show the variables that influence this sensitive marker of neurogenic disease.