ASSESSMENT OF MYOELECTRIC MANIFESTATIONS OF MUSCLE FATIGUE DURING REPETITIVE ISOMETRIC VOLUNTARY CONTRACTION IN BOYS

Aim: To study the peculiarity of electromyography signal characteristics alternation using different sEMG parameters during repetitive voluntary isometric fatiguing contraction in adolescent boys. Materials and methods. 12 subjects with height 148.75 ± 10 cm; Mass 38.9 ± 7.9 kg; age –12 to 14 years were recruited. The sEMG signal alteration of external oblique, rectus abdominis, erector spinae muscles during a fatiguing plank were analyzed. A separate one-way repeated measures ANOVA was used to test the statistical significance of task time and electromyography parameters of the global core muscle in the pre-, duringand post-fatigue plank test. Oneway Friedman ANOVA was applied for Shapiro-Wilk p < 0.05. The Pearson product-moment correlation coefficient with bivariate linear regressions analysis was performed between the pre-pre fatigue and post-post fatigue amplitude mean and standard deviation values. The Spearman correlation coefficient between amplitude and endurance time both in the preand post-fatigue state was conducted. Results. The mean value of rectified amplitude increased (p < 0.05) for all muscles, the standard deviation of amplitude and total spectral power increased significantly (p < 0.05) for all muscles except the erector spinae muscle (p > 0.05). The power at normalized low frequency significantly changed (p = 0.05) in the erector spinae muscle. A significant change in normalized low frequency for agonist/synergist (p = 0.02) and agonist/antagonist muscles (p = 0.05) was observed. The average amplitude value had a significant positive and linear relationship with the amplitude variability both in the preto post-fatigue state, except the erector spinae muscle. The time to task failure was not correlated (p > 0.05) with the sEMG amplitude. Conclusions. Increased sEMG amplitude resulted mainly from rapid additional motor unit recruitment and rate coding during muscle fatigue. The reduction of conduction velocity might affect the spectral power with a spectral shift towards low-frequency. Increased variability, agonist/antagonist co-activity during fatiguing contraction might extend the holding time. The postural fatiguing task/plank increases multiarticular joint function by involving several joints and muscles, increases variability in the contribution of synergist muscles. This factor provides an intuitive explanation about the absence of a relationship between endurance time and sEMG amplitude changes.


Introduction
The Surface Electromyography (sEMG) based measurement of "myoelectric manifestations of muscle fatigue" is meaningless if the receiver has a lower potential to understand the reliability and the information contained in the electrophysiological signals viz. muscle fiber propagation velocity, sEMG amplitude or frequency spectrum etc. as it also required technological proficiency. Significant gap between engineers and exercise physiologist/physical education teachers often observed in terms of barriers in sEMG signal interpretation is concern. The gap resulted mainly from nonfamiliar language and lack of technical background related to the application, signal processing and information extraction algorithms. Lack of knowledge in mathematics and biophysics is a significant educational barrier among physical education teachers often make it challenging for basic signal interpretation (Campanini, Disselhorst-Klug, Rymer, & Merletti, 2020;Felici & Del Vecchio, 2020).
The measurements of muscle fatigue usually performed during isometric contraction with a constant force, which is considered the "bench-test" condition. The "myoelectric manifestations of muscle fatigue" can be defined as the inability to maintain a given isometric contraction level and it is related to the endurance time with a relative sEMG features alteration from the very beginning of the contraction (Campanini, Disselhorst-Klug, Rymer, & Merletti, 2020). The retention time of the plank task could reliably estimate isometric core muscle endurance and test core stability in children (Boyer, Tremblay, Saunders, McFarlane, Borghese, Lloyd, & Longmuir, 2013). First time De Blaiser et al. (2018) reported that the plank test is valid and reliable to measuring global core muscle fatigue. They also stated that the past research studies intended to measure the validity and reliability of plank test was implemented sEMG as a reference technique designed to evaluate the parameters that failed to represent muscle fatigue.
Objectives: Despite the importance, few studies on exercise induced muscle fatigue in children and adolescents are available that makes it difficult to understand the neuromuscular components of pediatric muscle fatigue. Patikas et al. (2018) reported in their systematic review that previously cited research articles associated with exhaustive sub/maximal contraction induced muscle fatigue in children and adolescents revealed the requirement of more research in pediatric fatigue. The most study used only the normalized Root Mean Square value for muscle fatigue analysis. But the sEMG signal waveform cannot provide a valid conclusion about 'Muscle Fatigue' if we use a single sEMG parameter for a heterogeneous muscle group (Duchêne & Goubel, 1990). To our best information about existing literature on sEMG based muscle fatigue assessment, there are no such studies which evaluated myoelectric manifestation of core muscle fatigue extensively in children using exhaustive Plank test. Therefore we aimed to study the peculiarity of electromyography signal characteristics alternation using different sEMG parametres during repetitive voluntary isometric fatiguing contraction in adolescent boys.

Hypotheses:
Based on the analysis of previous sEMG fatigue studies we hypothesized that: The mean amplitude value of the sEMG signal was expected to have Gaussian probability distribution and it may show a significant positive and linear relation with absolute variability of sEMG amplitude. Pearson product-moment correlation coefficient (R). sEMG undergo non-uniform changes with increase variability (phase shift) during isometric fatiguing contractions.
3. H 0 : r = 0. H 1 : Significant Spearman correlation coefficient (r) between ARV amplitude with Endurance time both in pre and post fatigue state was expected.

Study participants
Sample size calculation: The sample size (n) was calculated based on the Average Rectified Value (ARV) of the Rectus abdominis muscle activity acquired in a pilot study with three children (Silva et al. 2020). We used ARV for sample size calculation because ARV is significantly less variable and reliable when measuring the core muscle activity (Hibbs et al. 2011). We included the pilot data in the main data set. The sample size was calculated using the G*Power (Version 3.1.9.2, Kiel University, Germany) application, for 80% power at an alpha level of 0.05. This calculation provided a sample size of 12 for this study.
Inclusion criteria: Prior to the collection of data, the children were requested to recognize their preferred writing hand, which was contemplated their dominant arm. All children were right-hand dominant (Silva et al., 2020). We collected the sEMG data of core muscle only from the right ISSN 1993-7989 (print). ISSN 1993-7997 (online). Теорія та методика фізичного виховання. Том 21, № 1 side in subjects with dominant right-hand for fatigue analysis because: 1. The fatigue indexes are significantly sensitive and evident in the right than left-handed subjects. The fatigue indexes are also less apparent in left than in right-handed subjects (Merletti, De Luca, & Sathyan, 1994). 2. The adaptation of the Motor Unit (MU) pool in the dominant muscle due to daily preferential use allows for more effective force production at low firing rates by increase the percentage of Type-I fibers resulting in twitch fusion at lower MU firing rates (Adam, De Luca, & Erim, 1998), although the previous study on hip-abductor fatigability reported non-significant results between dominant and nondominant side (Jacobs, Uhl, Seeley, Sterling, & Goodrich, 2005).
No previous history of Neuro/Myo pathological disorders and Postural Spinal deformities were reported during data collection. Those boys were clinically tested in (RKVM)-Sharada Balgram Health facility. : Full written advice about the possible risks and discomfort associated with the study was given to the children and local guardian/school principal, they signed the written informed consent form. The study was approved previously by the Departmental Research ethics board of Lakshmibai National Institute of Physical education, Gwalior, India and was conducted following the ethical principles for human research proposed in the Helsinki Declaration.

Study organization
Fatiguing Plank Protocol: Each child received verbal and visual instructions about the correct posture to perform plank to ensure comfort and familiarity. It may provide better sEMG signal quality during data collection. The Traditional plank protocol used in this study: face lied down and fists on the surface/floor, feet were placed at shoulder width apart, the spine and pelvis in a neutral position. The space between elbows were also shoulder width apart just below the glenohumeral joint. Lift the body up on the forearms and toes (Cortell-Tormo, García-Jaén, Chulvi-Medrano, Hernández-Sánchez, Lucas-Cuevas, & Tortosa-Martínez, 2017; Schoenfeld, Contreras, Tiryaki-Sonmez, Willardson, & Fontana, 2014;Bohannon, Steffl, Glenney, Green, Cashwell, Prajerova, & Bunn, 2018). Participants maintained the proper plank position and throughout the whole period we recorded the sEMG activity of the respective muscles until the test terminated, if the participants could not hold/continue with the correct position because of fatigue and pain. Maximum time limits were recorded with a stopwatch. In summary the children were asked to perform 5 consecutive Surya Namashkara with self-reported pace for warming up, followed by 5 mins. rest. After 5 min. rest they completed three planks (i.e. pre-fatigue (1st plank), during (2nd plank), and post fatigue (3rd plank) with a resting period of 3 mins. in between every plank. Strong verbal encouragements were given to make them motivated to hold a plank to their maximum potential limit. The 3 min rest or interval periods were given between every plank, because: Previous study reported that sufficient recovery from attenuated motoneuron excitability of the muscles induced by low/submaximal isometric fatiguing contraction requires ≤ 240 sec., resting period (Heroux, Butler, Gandevia, Taylor, & Butler, 2016).
Instrument: Preparation of Skin -A single-use cottonwool ball with 60-70% alcohol-based solution (isopropyl al-cohol) was used for clean the surface of the skin (hair saved) for better conductance.
sEMG Parameters: 1. A. ARV-EMG: The addition of all the multiplied values of an arbitrary parameter X and the probability of these values decide the centroid position in Gaussian distribution. B. SD EMG: Mean-squared deviation of the arbitrary parameter X-the positive square root of the squared deviation of a random parameter X from its mean value denote the scattering distribution. The whole EMG signal was full-wave rectified, smoothed. 2. N.LF -Normalized low-frequency index was the ratio of the area under the curve in the range of 10-30 s (LF), to the area under the curve in the range of 2-30 s. LF = N.LF/(TSP-VLF)·100 %. 3. Total Spectral Power: The frequencies were evaluated using Discrete-Time Fourier series analysis, with graphs of power or amplitude ratio (Y-axis-Power µV 2 ) of the oscillation period (X-axis Hz/Sec.). Total Spectral Power (TSP) = High Frequency (HF) + Low Frequency (LF). sEMG signal processing: The EMG linear envelope and power spectral density (PSD) was calculated using Welch and Bartlett's averaged modified periodogram (Non-parametric) method with 1024 sample analysis, 50% overlapping windows. In the power spectral density, the sEMG frequency cut-off or the band-pass filter was 10-512 Hz. Further, the energy/power alteration at normalized Low-frequency band 10Hz-70Hz was taken for fatigue analysis. As the attenuation of mean power frequency in Spectral parameters from 70 Hz to 60 Hz is considered as the progression of myoelectric manifestation of fatigue. The mean power of the frequency below 70 Hz increases significantly during fatigue. The low frequency increases linearly during the progression of myoelectric manifestation of fatigue (Cardozo, Gonçalves, & Dolan, 2011). Automated Linear spectrum interpolation algorithm was used to remove ТМФВ, 2021, том 21, № 1  sinusoid interference frequency (cosine function) and timecorrelated power line noise contamination. Electrodes placement (Right Side dominance): EOlateral portion of rectus abdominis muscle (~15 cm lateral to the umbilicus), situated straight over the anterior superior iliac spine, midway between the crest and ribs with oblique or 45˚ angle inclination. RA-~2 cm lateral to the umbilicus. ES-~2 cm parallel from the midline of the spine over the muscle. Ground electrode was placed over the midline of the lumbosacral bony landmark (Fig. 1) (Silva et al., 2020;Criswell, 2011).

Statistical analysis
The "Consensus for Experimental Design in Electromyography-2020" (CEDE) project provided the recommendations and guidelines for recording, analysis, interpretation and specific applications of EMG, published in the Journal of electromyography and kinesiology (International Society of Electrophysiological Kinesiology) (Besomi et al., 2020) reported to use non-maximal voluntary isometric contraction induced non-normalized sEMG data for interpretation as no method available for sEMG normalization in study dealing with pediatric population. Therefore sEMG signals (non-normalized) were used for fatigue analysis as per the instructions given by Besomi et al. (2020) and Halaki et al. (2012). The relative sizable inter-individual variability in the global core muscle activation pattern would be due to the diversity of anatomical and fiber type distributions. Therefore LOG(10) transformation procedures were applied for all sEMG Parameters and the alteration of sEMG activity of these three muscles was analyzed separately (Falla & Farina, 2007) as between muscle activity could not possible to calculate with non-normalized sEMG data (Besomi et al., 2020).
For the Normality distribution assumption, the Shapiro-Wilk test was used and a significant (p ≤ 0.05) value would be considered as the departure from normality. Separate oneway repeated measures ANOVA was used to test the differences across the pre-, during and post fatigue with relative alteration of different sEMG parameters of EO, RA, ES muscles, N.LF ratio for RA (Agonist)/ ES (Antagonist), RA (Agonist)/ EO (Synergist) muscles and plank/endurance time (T lim ). For each ANOVA, if those parameters were found significantly distinguishable, Bonferroni corrected paired t-test was used to identify differences (Post Hoc) for pairwise comparison. In case the assumptions for parametric statistics were not  ISSN 1993-7989 (print). ISSN 1993-7997 (online). Теорія та методика фізичного виховання. Том 21, № 1 met, nonparametric one-way Friedman ANOVA was used and statistically significant values were further analyzed with Dunn-Bonferroni test for multiple pairwise comparisons. The Sphericity assumption was tested by Mauchly's test and a significant (p ≤ 0.05) value would be considered as sphericity violation. The sphericity violation was treated with an appropriate degree of freedom (df) correction by applying either Greenhouse-Geisser (ε < 0.75) or Huynh-Feldt (ε > 0.75) correction. The level of significance was set at p ≤ 0.05. η p 2 was used as the effect size index. The values of η p 2 were interpreted as small effect (d = 0.20), medium effect (d = 0.50) and large effect (d = 0.80). For nonparametric Friedman ANOVA, Kendall's W coefficient was used for effect size, assuming the value from 0 (no relationship) to 1 (perfect relationship) (Silva et al. 2020). The Pearson product-moment correlation coefficient (R) was used in normally distributed (LOG(10)-transformed, Shapiro-Wilk p > 0.05) data. The R value between an independent and dependent parameter was tested with bivariate linear regressions analysis (Verma, 2016) and it was performed between the Pre-Pre fatigue and Post-Post fatigue LOG(10) SD EMG-(µV) with LOG(10) ARV EMG-(µV) of EO, RA, ES muscle separately. The Spearman correlation coefficient (r) of ARV EMG (μV) with Endurance time (T lim -Sec.) of EO, RA and ES muscle were applied for both in pre and post fatigue.

Results
The One-way Repeated ANOVA revealed a significant difference in the selected sEMG Parameters (p≤0.05) with low to large effect sizes (η p 2 ) ( Table 1). Larger effect sizes in selected sEMG Parameters were observed when compared with the baseline in the muscles with steeper fatigue level during repetitive fatiguing plank.
The predictive Linear regression model reveled (Fig. 4) Continuous sEMG recordings for the entire time to task failure during three consecutive plank (P1, P2 and P3 respectively) holding test (with 3 minutes interval in between). A. sEMG signals were full-wave rectified, smoothed. B. Power spectral density plots were obtained from EO, RA, ES muscles during static fatiguing plank task. The energy changed only at normalized Low-frequency band of 10Hz-70Hz was taken for fatigue analysis. Discussion of Hypotheses: According to the research hypotheses, some of the most important observations were that during sustained fatiguing contraction/plank the sEMG amplitude and the Total Spectral Power of the signal increased rapidly ( Fig. 2A), power spectra intensified/shifted towards low frequency power spectra (inflation in low frequency band) (Fig. 2B). The positive linear relationship between average amplitude values with amplitude standard deviations of the sEMG signal was altered by fatigue (Fig. 4). The nonsignificant relationship between the time to task failure with sEMG amplitude was influenced by complex multiarticular joint function.
Increased sEMG amplitude (Fig. 2 A, Table 1) resulted from rapid additional motor unit recruitment and rate coding. During fatiguing contraction, additional motor units recruited progressively as per Henneman's sizes principle. Firing off these motor units interfere and briefly inhibit the firing rate of previously recruited motor units, which might increase the variability (Table 1).
The concept of stretch reflex inflation during a postural fatiguing task is based on some basic theory: 1. The rapidly deteriorating in the inhibition of presynaptic transmission and feedback modulation by Ia afferents group. 2. Active dendritic actions modulated by descending pathway and supraspinal drive 3. Increase in motoneuronal α-γ coactivation. More pronounced postsynaptic than presynaptic activation mechanism was reported during fatigue if the pressor response absent. The motor neuron excitability during postural fatiguing task-induced pressor response activation was influenced by: 1. Group III-IV afferent input acquired by interneurons which excited the γ-motor neurons (Maluf & Enoka, 2005). 2. Variations in motor unit firing rate and recruitment pattern, increased synchronization, most importantly decreased muscle fiber conduction velocity, all of these factors might be acted as a compensatory mechanism during fatigue-induced force loss among muscles. It might change the shape of the action potential (Fig. 2), increase the magnitude (Mean) and Variation (SD) of ARV amplitude (Table 1). The primary effects of motor-unit synchronization were proposed during fatigue as: Augmentating the magnitude and variability of AVR EMG (intense-synchrony condition), intensification of low frequency in EMG power spectra, attenuation of the cancellation in overlapped positive and negative phases of action potentials (Yao, Fuglevand, & Enoka, 2000). Predominant muscle fiber-type (Type-I, 57-62%) and recruitment strategies (viz. increased motor-unit synchronization and decreased propagation velocity) of motor units might also be responsible for sEMG Spectral power alteration and amplitude augmentation. The motor units of type I (low threshold) muscle fiber generates action potentials with higher power at the lower frequency during fatiguing contraction. Both the myoelectric signal of muscle tissue and surface EMG electrodes act as low-pass filters, rapid energy of the signal travel through the tissue when the lower frequency intensifies or prominent leftward shift (Fig. 2) in signal eventuated in the power spectrum during fatigue (Fig. 2B) (Hagberg, 1981). Ischemia resulted from hemodynamic occlusion therefore anaerobic glycolysis during isometric contraction increases the production of lactate, pyruvate, inorganic Pi, resulted in intramuscular pH and conduction velocity to reduce. Studies reported lacking the significance of lactate production during fatiguing contraction proposed extracellular K+ accumulation and Na+ depletion in extracellular space could be the possible reason for sEMG signal alteration during fatigue (Linssen, Stegeman, Joosten, Binkhorst, Merks, Ter Laak, & Notermans, 1991).
Relatively less prominent antagonist muscle activity ( Fig. 2A) during the positional task was controlled by spinal interneurons, which further influenced by descending pathways to enhance the stability at the multiarticular joint level. The concurrent augmentation of agonist/synergistic (p = 0.02), agonist/antagonist (p = 0.05) N. LF ratio (Table 1) during fatigue resulted from either supraspinal descending drive or differentiated motor neuronal pool. The changes in excitability of motor neuronal pool in antagonist muscle usually observed with a substitute spinal pathway of disynaptic reciprocal inhibition from muscle's spindle afferents to the motor neurons as fatigue developed. An increase in sEMG amplitude (p = 0.04) of antagonist muscle stipulated a net increase in excitatory synaptic input to the spinal motor neurons and inherent characteristics of supraspinal mechanisms modulated by Ia presynaptic inhibition which further influenced different peripheral inputs parallelly as fatigue progressed. Researchers often reveal doubts about the reliability of antagonist sEMG (Low ICC in ES muscle) (Table 1). But with strict sEMG recordings protocol, it may produce an acceptable and meaningful antagonist coactivation even though the signal does not dispense a precise estimate of its mechanical contribution (Duchateau & Baudry, 2014). The modulation of antagonist coactivation during fatiguing contraction might extend the T lim (p = 0.56) (Table 1) (Shrawan et al., 1996). Motor units that innervate the number of muscle fibers reveal a Non-Gaussian distribution in sEMG Spectral power, uniquely the LOG(10)TSP showed unique alteration of kurtosis pattern among three muscles across the three consecutive fatiguing plank also might extend the T lim (Fig. 3). The failure to detect any statistically significant difference (p > 0.05) in sEMG parameters (Table 1) during fatiguing contraction might falsely be concluded as no specific alteration of rate coding and motor-unit activity. It was also possible that sEMG amplitude or power spectral frequency might show a lower potential to identify small changes efficiently of the neural drive against time of muscles with multiarticular joints functions, which further affected by signal contamination due to volume conduction, cross talk and electromagnetic interferences (Farina, Arendt-Nielsen, Merletti, & Graven-Nielsen, 2004). Although more repetitions with less resting or interval periods during standardized exercise (Kapkan, Khudolii, & Bartík, 2019) may increase the chance for significant sEMG signal alteration due to fatigue at this age group of this present study. Further studies where the children are involved in different exercise interventions may use different sEMG parameters with cautions to observe neural recruitment strategies alteration, overload/muscle fatigue level and training progression.
The average amplitude values of the sEMG signals usually have Gaussian probability distribution, often show a linear relationship with sEMG amplitude standard deviations of the signals. We also observed the average sEMG amplitude value produced a steeper and linear relationship with the amplitude variability with the relative alteration of the relationship from pre to post fatigue state of three core muscle (Fig. 4). Henneman's size principle stated small motor units (low recruitment threshold) have initial fatigue response, recruited to fire at low force level, shows small variability, where the larger motor units (high recruitment threshold) recruited at higher fatigue level indicated large variability. The subjects with larger variability showed a longer time to task capacity with more heterogeneous muscle activation pattern and smaller localized muscle fatigue. The neuronal mean firing rate alters sinusoidally and variability in spike generation is proportionately related to the mean firing rate during fatiguing contraction (Stein, Gossen, & Jones, 2005).
The non-significant (p > 0.05) Spearman correlation coefficient between the time to task failure with sEMG amplitude might influenced by multiarticular joint function, volume conduction, cross talk by the synergistic muscles or electrical interferences (Boyas, et al., 2009).

Limitations
Due to the complexity of MVIC induced sEMG amplitude normalization protocol for core muscle, clinicians usually avoid it, as it is also not recommended for the children less than 15 years age (Nicholson, 2000). The electromagnetic interferences, crosstalk by surrounding muscles could never be replaced completely but tried to reduce it with high signalto-noise ratio in sEMG setting.

Conclusions
Increased sEMG amplitude resulted from rapid additional motor unit recruitment and rate coding while performing the fatiguing plank task. Variation in motor unit firing rate and recruitment pattern, increased synchronization and deterioration in the muscle fiber conduction velocity, all of these factors influenced the shape of action potential and made the sEMG amplitude increase during fatigue. Recruitment strategies of motor units and predominant I-type muscle fiber, these two factors caused sEMG Spectral power to change and spectral shift towards low-frequency band during fatigue. The longer time to task or Endurance capacity was observed by heterogeneous muscle activity during isometric contraction. Postural fatiguing task/plank increases multiarticular joint function by involving several joints and muscles, increases variability in the contribution of synergist muscles. This factor provide an intuitive explanation about the absence of a relationship between endurance time and sEMG amplitude changes.