Variability of estimates of muscle fiber conduction velocity and surface EMG amplitude across subjects and processing intervals

Abstract

The force developed by the human neuromuscular system can change very rapidly (15-50 ms). When processing EMG signals for inferring neural control strategies, it is therefore necessary to extract estimates from short time intervals. In this study, we investigate the relation between joint torque and estimates of average muscle fibre conduction velocity (MFCV) and amplitude (RMS) from surface EMG signals, when varying the duration of the processing interval. Moreover, we assessed the inter-subject variability in RMS and MFCV estimates. Ten healthy subjects performed isometric linearly increasing ankle dorsiflexion contractions up to 70% MVC at a rate of 5%MVCs^-1. High-density EMG signals were recorded from the tibialis anterior muscle and MFCV and RMS were estimated in eight time-intervals ranging from 15 to 2000 ms. MFCV and RMS were significantly correlated with force in all subjects and when using all time-intervals (MFCV = 0.77 ± 0.07, RMS = 0.79 ± 0.06 (R²), Pearson-P < 0.01). The variability around the regression line for both MFCV and RMS estimates significantly increased when using intervals <100 ms (P < 0.001). However, the slope of the regression between EMG variables and force did not change with the duration of the interval (P < 0.001). Moreover, MFCV showed a substantially smaller variability across subjects in its relation to force than RMS [average coefficient of variation of regression slopes across all time intervals, 24.48 ± 1.51 (%), whilst for the RMS it was 56.65 ± 0.69 (%)]. These results indicate that estimates of MFCV and RMS as a function of joint torque are unbiased with respect to processing interval duration. Moreover, they reveal that estimates of MFCV are more consistent across subjects than EMG amplitude.

Keywords: EMG epoch length; EMG features; EMG inter-subject variability; EMG time frame; Motor unit recruitment.