50 a REF in the present study.
3.2-51
1a). However, those phases with assistance provided only the target trajectory of the REF and the linear actuator was operated to induce force control with assistance (Figure 3.2-1b; shaded area). For example, if a 50% assistance of the REF was to be provided with ramp-down duration of 2.5 s, the same amount of the REF was provided as an assistance with force increase duration of 2.5 s, based on the Eq. (1) (Figure 3.2-2). Thus, mechanically-assisted force control was conducted by tracking indirect visual guidance of the REF trajectory. The real-time force control feedback of 𝐹𝑡 was provided as a continuous dotted line concurrent with the target trajectory. Throughout the experiment, participants were asked to track and match the force control feedback to the target trajectory as accurate as possible and were prevented from anticipating the timing of providing mechanical assistance and visual guidance. One experimental task lasted about 60 s.
Experimental conditions
Within the subtask, the duration of ramp-down (the second phase of subtask) was set either 0.5 s, 2.5 s, or 5.0 s. Next, the following hold phase was required to hold the step-down magnitude of either 35.3 N or 23.5 N, which represents 75% REF or 50% REF, respectively. Therefore, the force release phase involved one of six different ramp-downs:
releasing force to 75% REF in 0.5 s (-23.4 N/s), 2.5 s (-4.7 N/s), and 5.0 s (-2.4 N/s);
releasing force to 50% REF in 0.5 s (-47 N/s), 2.5 s (-9.4 N/s), and 5.0 s (-4.7 N/s).
52
Procedure
The experiment consisted of three sequential sessions: (1) MVCs measurements for the BB and TB muscles; (2) task familiarization; and (3) main experiment of trajectory tracking task. After the measurement of the MVCs, each participant was familiarized with the experimental task by practicing it at least two trials. During the main experiment, each participant repeated twelve trials of the trajectory tracking task, and 5 min of rest was given between each trial. All experimental conditions were counterbalanced to minimize the predictability and carry-over effect. Total sessions lasted about 2.5 h.
Figure 3.2. (1) An example of the trajectory tracking task and (2) a representative force output from the linear actuator.
53
Measurements
Surface EMG was recorded from the two muscles acting on the elbow flexion: the BB and TB muscle. The surface of those muscles was palpated and scrubbed with alcohol, after which a pair of disposable surface electrodes (N-00-S, Ambu Inc., MD, USA) was attached on each surface in line with the direction of muscle fiber, following the SENIAM recommendations (Hermens et al., 2000). The distance between the two electrodes was 25 mm. Ground electrodes were attached on the right acromion and the head of radius.
The EMG signals were amplified using a Bio-amp ML 132 (ADInstruments; × 1,000), band pass filtered (15–500 Hz), and full-wave rectified. The measurements of the accelerometer were high-pass filtered with a cut-off frequency of 0.7 Hz to eliminate the effect of gravitational acceleration (Usui et al., 2015).The signals of the EMG, the force outputs measured from the two load cells, and the acceleration measured from the accelerometer were sampled by the A/D converter at 1 kHz, before being stored on a computer.
During MVC measurement session prior to the main experiment, participants performed at least three maximal contractions for each of the BB and TB muscles. One MVC trial lasted 7 s with 60 s of rest provided between the trials. The mean rectified amplitude over the middle 3 s of MVC trials was quantified for those muscles and used
54
as values for EMG normalization (%MVC). The maximal force during isometric elbow flexion was also obtained from the load cell connected between the level of the styloid process of right arm and the ground, simultaneously with the MVC trials of the BB muscle.
Data analysis
This study mainly analyzed data for the results of tracking two phases: force release and holding the step-down magnitude (ramp-down and hold phases) in the experimental task. The final 2 s of the data were discarded to collect steady-state data.
Averaged values of the repetitions were used for data analysis. Normalized muscle activity (%MVC), force output (N), and vertical perturbation (m/s2) were analyzed, and these measurements were used to derive applied parameters, as shown below. The data were processed using Labchart 7 and Excel (Microsoft, Redmond, WA, USA).
Relative EMG power. The EMG data measured from the BB were examined in the
frequency domain using Welch's averaged periodogram method with a window size of 2,048 points and 75% overlap. EMG power spectra were calculated from the rectified, band-pass filtered data. These were examined in a frequency range of 15-45 Hz, as related studies regarding the CMC have shown that EMG oscillation in the 15-45 Hz bandwidth is associated with motor function of a steady and precise motor output during visually guided isometric contraction (Chakarov et al., 2009; Kenway et al., 2016). The percentage
55 of the total power within the bandwidth was calculated.
Muscle co-contraction ratio. The relative involvements of the BB and TB muscles
during the isometric force control were estimated by Eq. (2.1) in Chapter 2.
Force variability. Fluctuations of force during the isometric force control were
quantified from the CV of 𝐹𝑚 force output (Eq. (3.2)) (Baweja et al., 2009; Tracy et al., 2007; Welsh et al., 2007).
Force variability (%) = SD of Fm
Mean of Fm×100 (3.2)
Overshoot ratio. The degree of overshoot, which may arise during force control
from the REF (top) to step-down magnitude, was quantified by normalizing the difference between the actual and target force changes. The maximum difference was adopted according to Eq. (2.3) in Chapter 2.
Peak perturbation. Peak perturbation (m/s2) during isometric elbow flexion was quantified as the maximum root mean square value of perturbation for every 10 samples within the initial 0.5 s of force release phase.
56
Statistical analysis
A three-way repeated measures ANOVA was conducted to investigate the effects of the assistance operation (assistance with indirect visual guidance and non-assistance with direct visual guidance), ramp duration (0.5 s, 2.5 s and 5.0 s) and step-down magnitude (75% REF and 50% REF) on the aforementioned variables. Post hoc analyses were conducted using the Bonferroni’s correction for multiple comparison adjustments.
SPSS Statistics 23.0 (IBM, Research Triangle Park, NC, USA) was used for the statistical analyses, and statistical significance was determined by α = 0.05. All values are reported as mean ± SD. Partial eta squared (η2) was provided as the effect size.
57
3.3 Results
Force output and muscle activity
The MVC force measured during isometric elbow flexion was 185.7 ± 48.1 N. Thus, the normalized force required for maintaining the REF (47 N) across participants was 26.8 ± 6.3% of the MVC force. Figure 3.3 demonstrates representative outputs of manual force control (𝐹𝑚) and corresponding muscle activity during the trajectory tracking task.
The force was successfully released from the REF and held under the conditions of ramp duration and step-down magnitude as planned, independently of assistance operation.
Accordingly, the force output and activities of the BB and TB muscles showed patterns similar to each other. This study examined the last 3 s of these measurements to check whether holding step-down magnitude was biased by any experimental conditions.
The results of ANOVA yielded significant main effects of step-down magnitude for all of those parameters (force output: F (1,12) = 4505.44, p < 0.01, η2 = 0.99; BB: F (1,12)
= 17.46, p < 0.01, η2 = 0.59; TB: F (1,12) = 14.12, p < 0.01, η2 = 0.54), showing that force output and EMG amplitudes were higher while holding step-down magnitude of 75%
REF (force output: 34.9 ± 0.8 N; BB: 20.6 ± 10.0%; TB: 5.5 ± 3.1%) than holding step-down magnitude of 50% REF (force output: 23.2 ± 1.1 N; BB: 15.7 ± 7.5%; TB: 4.5 ± 2.4%). In contrast, the main effects of ramp duration and assistance operation were not
58
significant for either force output or EMG amplitudes of the BB and TB during the last seconds.
Figure 3.3. The representative patterns of force output of 𝐹𝑚 and rectified EMG amplitudes of the BB and TB, during force release and hold phases under the ramp duration of 0.5 s (averaged for n = 13). In these graphs, sampling frequency was set as 10Hz for convenient visualization.
59