Tracking motor units longitudinally across experimental sessions with high-density surface electromyography

Abstract

Key points: Classic motor unit (MU) recording and analysis methods do not allow the same MUs to be tracked across different experimental sessions, and therefore, there is limited experimental evidence on the adjustments in MU properties following training or during the progression of neuromuscular disorders. We propose a new processing method to track the same MUs across experimental sessions (separated by weeks) by using high-density surface electromyography. The application of the proposed method in two experiments showed that individual MUs can be identified reliably in measurements separated by weeks and that changes in properties of the tracked MUs across experimental sessions can be identified with high sensitivity. These results indicate that the behaviour and properties of the same MUs can be monitored across multiple testing sessions. The proposed method opens new possibilities in the understanding of adjustments in motor unit properties due to training interventions or the progression of pathologies.

Abstract: A new method is proposed for tracking individual motor units (MUs) across multiple experimental sessions on different days. The technique is based on a novel decomposition approach for high-density surface electromyography and was tested with two experimental studies for reliability and sensitivity. Experiment I (reliability): ten participants performed isometric knee extensions at 10, 30, 50 and 70% of their maximum voluntary contraction (MVC) force in three sessions, each separated by 1 week. Experiment II (sensitivity): seven participants performed 2 weeks of endurance training (cycling) and were tested pre-post intervention during isometric knee extensions at 10 and 30% MVC. The reliability (Experiment I) and sensitivity (Experiment II) of the measured MU properties were compared for the MUs tracked across sessions, with respect to all MUs identified in each session. In Experiment I, on average 38.3% and 40.1% of the identified MUs could be tracked across two sessions (1 and 2 weeks apart), for the vastus medialis and vastus lateralis, respectively. Moreover, the properties of the tracked MUs were more reliable across sessions than those of the full set of identified MUs (intra-class correlation coefficients ranged between 0.63-0.99 and 0.39-0.95, respectively). In Experiment II, ∼40% of the MUs could be tracked before and after the training intervention and training-induced changes in MU conduction velocity had an effect size of 2.1 (tracked MUs) and 1.5 (group of all identified motor units). These results show the possibility of monitoring MU properties longitudinally to document the effect of interventions or the progression of neuromuscular disorders.

© 2016 The Authors. The Journal of Physiology published by John Wiley & Sons Ltd on behalf of The Physiological Society.

Figures

Figure 1. Motor unit decomposition and tracking procedure

A, high‐density surface EMG signals (64 channels) were recorded from the vastus medialis (VM) and vastus lateralis (VL) muscles during a ramped isometric knee extension (50% of the maximum voluntary contraction (MVC)). The EMG signals were decomposed to reveal the firing activities of single motor units. A schematic representation of the task and motor unit (MU) recording methodology is shown in the left half of the figure. B, the procedure developed in the study was then used to identify two matched MUs between the first and the last session of experiment I. The cross‐correlation between the motor unit action potential profiles of the identified MUs was higher than 90%. Multichannel action potentials (59 bipolar channels) of the original (blue) and matched (red) MUs are shown to confirm their similar MU action potential shapes. Two matched MUs are being shown on the right side of the figure (1 for VM, up and 1 for VL, down). For clarity, MU action potentials inside the dashed boxes are zoomed in the right half of the figure. Those matched MUs had cross correlation coefficients > 0.9.

Figure 2. Motor unit tracking across sessions

A, multichannel surface action potentials of 3 different vastus medialis motor units (MUs) that were tracked across the three sessions. The cross correlation coefficients (CCCs) of the MU action potential profiles between the three sessions can be seen above. For the sake of clarity MU action potential matching is presented between two sessions only. MU action potentials extracted from the first session are presented in blue while matched action potentials from the second session are presented in red. B, discharge times of each matched MU during ramped contractions at 30% MVC during the 3 sessions, note the similarity of their recruitment and de‐recruitment thresholds.

Figure 3. Motor unit tracking and changes in conduction velocity

A, vastus medialis (VM) and vastus lateralis (VL) motor unit action potentials (MUAPs) that were identified by the tracking algorithm before (pre, red) and after (post, blue) the endurance intervention at 30% of the maximum voluntary contraction (MVC) force. Conduction velocity values can be seen above the MUAPs. B, cross‐correlation of the VM and VL MUAPs identified pre and post training. Cross‐correlation coefficients (CCCs) from tracked motor units can be seen above the matched MUAPs. Note the similarity in action potential shape for the tracked motor units despite the large increases in conduction velocity.

Figure 4. Changes in conduction velocity for tracked and total group of identified motor units

A, motor unit conduction velocity (CV) values from the vastus medialis (VM) at 30% of the maximum voluntary contraction (MVC) from n = 7 subjects, previously (PRE) and after (POST) an endurance training intervention. Left graph shows results obtained with tracked motor units, while right graph shows the results obtained using the total group of identified motor units (CV values were averaged per subject and compared PRE and POST intervention). The effect size and P values of the two procedures are shown in the lower right corner of all graphs. The red line depicts an example of one subject that showed an increase in CV of matched motor units (left), which is masked when using the total sample of identified motor units (right). B, matched (left) and total sample of identified (right) motor units (mean and 95% confidence interval), from the same subject depicted in A (red line). The 12 matched motor units from this subject show a clear intervention effect (left graph), which is not possible to distinguish when using all decomposed motor units (CV values are extracted from all the motor units decomposed pre and post intervention (two repetitions per session)).