Higher involvement of synergistic muscle than active muscles in ankle plantar flexion

Abstract

The aim of this study was to observe the activation characteristics of calf muscles in healthy adults during plantar flexion by a new nonlinear electromyography index. The linear indices of sEMG, including the root mean square (RMS), median frequency (MF), and nonlinear index, degree centrality (DC) of the calf muscles of ten healthy male participants, were tested in the resting state and plantar flexion position to analyse their characteristics. The RMS was not normally distributed, and the MF and DC values were normally distributed. Compared with those in the resting state, the RMS, MF and DC values of all the tested muscles were significantly greater (P < 0.05) during plantar flexion. Among all the muscles, the peroneus longus (PL) changed the most. The RMS increased from 3.14 ± 1.61 µV to 49.31 ± 21.81 µV, the MF increased from 134.07 ± 30.86 Hz to 203.55 ± 28.36 Hz, and the DC changed from 2.24 ± 0.38 to 3.33 ± 0.41. Two linear indices and a nonlinear index of the PL were significantly elevated during prone plantar flexion. We speculate that the nonlinear index, DC can reflect the degree of single muscle synergistic involvement, and may become a new indicator for assessing the state of single muscle controlled by nervous system in a movement involving multiple muscles.

Background

Flexible ankle‒foot mobility is accurately controlled and adjusted by calf muscles. Ankle‒foot dysfunctions usually lead to compensatory movement and secondary injuries such as falls^1,2. These dysfunctions often result from changes in the neuromuscular conditions of calf muscles. In the clinic, these dysfunctions are easily observed in various neurological diseases, such as poststroke hemiplegia³cerebral palsy⁴and spinal cord injuries⁵. Therefore, recovering the motor function of calf muscles is very important for patients with neurological diseases.

Muscle synergy analysis provides a robust framework for dissecting the neural control mechanisms underpinning the function of calf muscles. Specifically, muscle synergy represents a neural strategy wherein the central nervous system simplifies motor control via the recruitment of coordinated muscle activation patterns^6,7. This modular approach reduces high-dimensional motor control problems to low-dimensional synergistic patterns by decomposing multi-muscle electromyographic signals. Researchers can extract the activation profiles of synergistic muscle groups, thereby quantify their capacity for coordinated control.

Surface electromyography (sEMG) reflects muscle activation by linear indices such as the root mean square (RMS), average electromyography (AEMG), and median frequency (MF)^8–10. Once indices exist to accurately reflect the characteristics of multimuscle activation coordination, clues of how the brain organizes movements can be identified. Recent studies have suggested that nonlinear network analysis of sEMG can better reflect the synergistic relationships of muscles in the process of motor control^11,12.

In previous studies, our team applied nonlinear indices, including degree centrality (DC), average interlayer mutual information, and average edge overlap, in muscle coordination analysis and reported that nonlinear indices have the potential to distinguish neuromuscular control patterns^13,14. The DC is a graph-theoretic metric based on functional network analysis, which is typically used to quantify the number of edges connected to a node and reflect a node’s integral contribution to the overall structure. The larger the DC of a node, the more important the node is. It is used mainly in the field of brain function and relatively rarely in the field of muscle assessment^15–17. In muscle network analysis, the DC can reflect the degree of active activation of different muscles and their involvement in a particular activity. A larger DC of a muscle was associated with greater involvement and contribution to multimuscle coordination¹⁸. Compared with the DC values in the healthy population, DC values in patients that are either excessively large or small suggest abnormal muscle synergy.

In this study, we intended to assess the activity of the calf muscles during plantar flexion in healthy subjects by using the linear indices RMS and MF and the nonlinear index DC of sEMG. The similarities and differences between the linear and nonlinear indices could be used to investigate the activation of the calf muscles during plantar flexion in healthy subjects and to further analyse the potential application of DC in the assessment of muscle function.

Methods

This was a cross-sectional observational study. This study was approved by the Ethics Review Committee of Suzhou Hospital Affiliated with Nanjing Medical University, and the study registration number is ChiCTR2100055090.

Participants

Ten healthy male volunteers, with an average age of 54.0 ± 8.1 years, an average weight of 68.3 ± 8.2 kg, and an average height of 167.9 ± 5.3 cm, were recruited for this study. All the subjects were right-handed, in good health, with no history of lower limb injury or nervous system or musculoskeletal disease and no restrictions on ankle dorsiflexion. All the subjects voluntarily participated in this experiment and signed an informed consent form before the experiment, informing them of the experimental procedure and purpose.

Experimental procedures

Before the test, the participants were instructed by the tester to familiarize themselves with the testing environment and procedures. During the formal test, participants lay prone on the examination table with their feet off the edge of the table, completely relaxed and in a neutral position (knees fully extended and ankles neutral with no rotation). A previous study recommended testing the plantar flexor muscles in the prone position, as this position was better at avoiding synergistic activation of the flexor muscles and reducing internal and external rotation of the ankle joint¹⁹. This study employed a standardised operating procedure: participants were first guided through pre-adaptive training via verbal instructions (the instruction protocol: “Slowly flex the foot toward the lower leg to the maximum angle, maintain this position for 10 seconds with maximum force”), ensuring they mastered the movement technique. All participants completed three movement simulation training sessions prior to the formal test, with researchers providing real-time corrections for joint angle deviations to ensure that the ankle joint would remain in a neutral position throughout the test. This standardization ensured uniform joint movement angles across participants and avoided muscle activity fluctuations caused by postural differences. The participants rested for 1 min between all tests, and the test was repeated three times. A total of three tests were averaged.

For the experiments, the tibialis anterior (TA), peroneus long (PL), peroneus brevis (PB), soleus (SOL), gastrocnemius medial (MG), and gastrocnemius lateral (LG) were selected from the left and right sides of the subject. The TrignoTM wireless surface EMG acquisition system from Delsys, USA, was used to record the sEMG signal. In accordance with the relevant European Union recommendations for sEMG signal electrode adhesion²⁰each muscle site was located, the relevant site was shaved, and the hair was polished with abrasive cream and wiped with alcohol. After the skin was dry, the sEMG electrodes were pasted sequentially in the direction of the muscle fibres onto the highest surface elevation of the muscle belly of the 6 groups of muscles on both sides of the subject. The sampling frequency of the device was set to 1926 Hz.

Data Analyses

The sEMG signals were preprocessed, and indices were extracted using MATLAB R2020a (MathWorks, Natick, MA, USA). To address signal instability at the start and end of each trial, an 8-second intermediate segment was selected for subsequent analysis. To filter the sEMG signals, a fourth-order Butterworth bandpass filter with a frequency range of 20–500 Hz was employed. To extract the assessment indices, we adopted a sliding window approach with 1000 sample points and an overlap of 250 sample points. The average of all windows’ assessment indices served as the outcome for each individual trial.

Linear indices

Linear time‒frequency domain indices were extracted from six muscles on both sides of the subjects. Under the assumption that the collected sEMG signals are represented as , , where the signal length is denoted. The RMS reflects changes in signal amplitude over time, serving as an effective measure of neural discharge, with the calculation formula as follows:

1

The median frequency (MF) is employed to characterize changes in the spectral features of sEMG signals, and its calculation formula is as follows:

2

where the power spectrum refers to , the range of frequency of the power spectrum between and .

The DC

The methodology for constructing functional muscle networks in this study builds upon the recursive network proposed by Deniz Eroglu in 2018, with extensions and adaptations applied²¹. On the basis of the theory of phase space reconstruction, an adjacency matrix for a single-channel sEMG signal is initially constructed, resulting in a recurrence network (RN) for each individual muscle. Each RN is subsequently treated as a node, and interlayer combinations are formed by considering the mutual information between nodes as the weight for connecting two nodes. This process yields a weighted network (WN) representing interconnections between muscles. The detailed process is illustrated in Fig. 1.

Fig. 1.

Schematic of the WN construction.

The DC of the node is defined by the formula²²:

3

where represents the weight of the edge connecting nodes and and where denotes the set of nodes. The average DC of all the time nodes is taken as the final DC of an individual muscle. Higher DC values indicate a more crucial position for the node within the network²³.

In this study, the mean values of the RMS, MF and DC on the left and right sides of healthy subjects were taken as the results of the muscles of the calf. In addition, the DC were sorted to observe the change in muscle involvement during sorting after plantar flexion.

Statistical Analyses

The statistical package SPSS version 26.0 (IBM Corp, Armonk, NY) was used for data analysis. Normally distributed data are represented by the mean ± standard deviation(); non-normally distributed data are represented by the median and interquartile range. When comparing data, if the data was normally distributed, a paired t-test was used; if the data was not normally distributed, the rank sum test was used for comparisons between two conditions. P < 0.05 was considered statistically significant.

Results

The RMS

The RMS values exhibited a nonnormal distribution. With respect to plantar flexion, the RMSs of the TA, LG, MG, FDL, SOL, PL and PB were significantly greater than those at rest (P < 0.05) (see Table 1).

Table 1.

The RMS and MF of sEMG between the two conditions.

Variables	Rest state	Plantar flexion
Tibial Anterior
RMS(µV)	2.64 ± 0.39	6.17 ± 2.60a
MF (Hz)	151.81 ± 23.91	183.22 ± 23.48a
Lateral Gastrocnemius
RMS(µV)	3.35 ± 1.15	26.79 ± 18.26a
MF (Hz)	137.11 ± 31.58	186.82 ± 27.36a
Medial Gastrocnemius
RMS(µV)	2.89(2.60, 3.49)	30.47(25.51, 47.67)a
MF (Hz)	153.85 ± 26.92	192.24 ± 22.71a
Soleus
RMS(µV)	4.09 ± 1.19	28.10(19.18, 35.73)a
MF (Hz)	127.52 ± 37.98	176.51 ± 27.96a
Peroneal Longus
RMS(µV)	2.51(2.26, 3.28)	49.31 ± 21.81a
MF (Hz)	134.07 ± 30.86	203.55 ± 28.36a
Peroneus Brevis
RMS(µV)	3.78 ± 0.90	38.79 ± 29.11a
MF (Hz)	150.59 ± 37.31	178.72 ± 22.14a

^aSignificant difference between rest state and plantar flexion.

The MF

The MF values exhibited a normal distribution. In plantar flexion, the MF of the TA, LG, MG, FDL, SOL, PL and PB increased compared with those at rest (P < 0.05) (see Table 1).

The DC

The DC values exhibited a normal distribution. In plantar flexion, the DC values of the TA, LG, MG, FDL, SOL, PL and PB were significantly greater than those at rest (P < 0.05) (see Table 2).

Table 2.

The DC in the two conditions.

Muscles	Rest state	Plantar flexion
Tibial Anterior	2.13 ± 0.20	2.89 ± 0.22a
Lateral Gastrocnemius	2.19 ± 0.30	3.29 ± 0.40a
Medial Gastrocnemius	2.25 ± 0.41	3.31 ± 0.42a
Soleus	2.33 ± 0.40	3.30 ± 0.37a
Peroneal Longus	2.24 ± 0.38	3.33 ± 0.41a
Peroneus Brevis	2.25 ± 0.33	3.22 ± 0.34a

^aSignificant difference between rest state and plantar flexion.

The DC values of the 6 muscles were sorted from largest to smallest. The DC values of the lower limb muscles at rest were ranked as SOL, PB, MG, PL, LG, and TA. The DC values of the lower limb muscles in plantar flexion were ranked as PL, MG, SOL, LG, PB, and TA (see Fig. 2).

Fig. 2.

The differences in the ranking of the DC between the two conditions.

Discussion

To investigate the potential application of DC in the assessment of muscle motor function, this study used RMS, MF, and DC to assess the calf muscles at rest and during plantar flexion in healthy adults. Similarities and differences between the linear and nonlinear indices were observed. Two linear indices and one nonlinear index of the PL were significantly elevated during prone plantar flexion, thus prone knee extension may be the optimal position for training the PL. DC may reflect the altered synergistic involvement of individual muscles in the muscle network.

The RMS

The results demonstrated a significant increase in the RMS of the tested muscles on both sides during plantar flexion. Specifically, the RMS for PL, PB, and MG exhibited greater increases, whereas the TA presented the smallest increase. Notably, the values of RMS are distributed unnormally. As the RMS is often used for reflecting degrees of muscle recruitment activation^24–26this outcome suggests greater activation of the lateral muscle group than of the active plantar flexor muscles, contradicting the established understanding that the movement of plantar flexion primarily depends on the agonistic muscles at the back of the calf, whereas the lateral muscles play a secondary supporting role^27,28. We hypothesize that the results of this study occur because active plantar-flexor muscles insert at the medial aspect of the calf and produce ankle inversion; thus, the PL and PB, in addition to assisting with plantar flexion, act as neutralizing muscles to counteract the redundant inversion effect caused by the plantar-flexor muscle group, thereby maintaining the ankle neutral position. Therefore, the increased recruitment of PL and PB during plantar flexion with the feet in the neutral position seems reasonable. The TA, as the antagonist muscle, also has a small increase in the RMS value. This is because during plantar flexion, the central nervous system controls the antagonist muscle to coordinate the relaxation while also recruiting a certain number of motor units to participate in activity. This primarily aims to maintain joint stability and the rational distribution of loads borne by periarticular tissues to protect the articular tissues²⁹.

These findings indicate that muscles of different sides or functions have different RMS values. In addition, neutralizing muscles may have higher RMS values than agonistic muscles.

The MF

The results revealed that the MF of all the tested muscles increased during plantar flexion compared with the rest state. This finding indicated that the motor units of all muscles were activated³⁰. In the static state of maximum plantar flexion in the prone position, all the muscles, including the posterior agonistic muscles, lateral fixation or neutralizing muscles, and anterior antagonist muscles, are in isometric contraction. This finding suggested that the high activation state shown by RMS and MF resulted from the increase in muscle recruitment and synchronization during the isometric contraction of each calf muscle.

The DC

A typical application of DC lies in measuring the quantity of direct connections to an assigned node. The DC can assess a node’s contribution to the overall structure or organizational activity by calculating network density. This measurement indicates the extent to which a node influences the entire brain and integrates information from brain regions separated by functional boundaries^23,31. In muscle network analysis, the DC reflects the activation levels of different muscles during a specific activity. Higher values of the DC of a muscle are associated with more involvement and contribution to multi-muscle coordination¹⁸.

The results revealed that the DC values of all muscles was greater in the plantar flexion state than in the resting state. This result showed that all muscles of the calf had increased synergy with the surrounding muscles during plantar flexion. Moreover, the DC ranking results suggested that the ranking of PL improved greatly during the test. This result may indicate that nonlinear network indices, such as DC, may have a potential function in determining the heterogeneity of multimuscle load-sharing strategies during movements in a new way. This study demonstrated that plantar flexion in the prone position could significantly activate the PL. A previous study demonstrated that walking on a medial inclined ramp could effectively activate the PL³². Although the relative effectiveness of these two modalities in PL training remains to be elucidated, walking on the medial inclined ramp is more demanding for subjects and carries a greater risk of falling than the prone knee extension posture, a safer position.

Differences in DC with linear indices

The results revealed that the changes in the DC value of all muscles during plantar flexion were consistent with the linear indices (see Figure 3). However, these indices represent the activation of each muscle of the calf from different perspectives: increased RMS and MF reflect increased activation and frequency of firing of multiple muscle fibres within the subject muscle, whereas DC reflects increased intermuscular connectivity. The results revealed a significant improvement in both intramuscular muscle fibres and intermuscular associations during the prone plantar-flexion manoeuvre.

Fig. 3.

Changes in three indices for all muscles in two conditions.

In addition, our findings showed that the DC values were normally distributed, suggesting that DC could be used to compare the degrees of activation of muscles with varying dimensions. The RMS and MF are composite indices influenced by various factors, including neural activation conditions, muscle fibre characteristics, muscle cross-sectional area, and subcutaneous tissue thickness³³. However, DC reflects the degree of intermuscular association, independent of the muscles’ intrinsic firing rates, which can better minimize the effect of individualized differences and could be used to compare the degrees of activation of muscles with varying dimensions.

DC quantifies the degree to which a single muscle is connected to the rest of the muscle network, reflecting the degree of synergistic muscle involvement. Various studies have shown that differences in muscle synergy are attributed to neural control^34–36and patients with neurological disorders such as stroke and cerebral palsy have impaired muscle synergy^37,38. A recent study further revealed that in hemiplegic patients, the DC values of the affected-side MG were significantly lower than those in healthy controls. Conversely, DC values of synergistic muscle groups on the unaffected side which compensate for the affected side were significantly elevated. These findings indicate that neuro-muscular control adaptations occur in the unaffected side of hemiplegic patients, and DC can effectively reflect such changes¹⁸.

Study Limitations

There are several limitations of this study. This study explored only healthy male groups, and only a small sample size was selected. Further studies can explore whether DC can meaningfully change as the contraction level increases, apply it in muscle assessment in patients with motor dysfunction and compare it with other indices.

Conclusion

This study investigated the characteristics of two linear indices and one nonlinear index of calf muscles in healthy adults during plantar flexion. Both RMS and DC showed greater activation of the PL during prone plantar flexion. The DC further revealed higher synchrony between the PL and other muscles, which may serve as a novel index for evaluating neural control of multi-muscle movements.

Author contributions

YH conceived and designed the study. YH, final approval of the version to be submitted. YX drafted the article. JL: acquisition and calculation of data. XJ analysed and interpreted the data. All authors have read and approved the final work in the author contribution statement.

Funding

This study was supported by the Science and Technology Program of Suzhou (SKY2021053 & SKY2022186).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

The experimental procedures were approved by the Ethics Committee of Suzhou Municipal Hospital (KL901116 and K-2022-161-K01) and were in accordance with the Declaration of Helsinki.

Competing interests

The authors declare no competing interests.

Written informed consent was obtained from the patient before this manuscript was reviewed for both case description and approval for the publication of this manuscript.