The Influence of Lower Limb Muscle Selection on Synergy Analysis during Running ^†

Abstract

Muscle synergies have emerged as a novel framework that offers a deeper understanding of the physiological basis of motor control and have been extensively applied in various fields. This study aims to investigate muscle synergies, involving the selection of the muscles of the lower limb while running, including the deep muscles. The muscle synergies were identified from the 23 muscle activations generated by the OpenSim simulation and experimented EMG, respectively, and these outcomes were compared. Our results revealed an additional synergy required to achieve a comparable variability accounting for (VAF) levels when considering 23 muscles compared to 10 muscles. This study highlights that the potential contribution of deep muscles to explaining the muscle synergies involved in running. These findings provide valuable insights for a comprehensive understanding of muscle synergy during running and the neuromotor control of human locomotion. Further optimization of the experimental protocols and extraction methods is necessary to enhance the results of muscle synergy analyses and promote their broader application across various fields in the future.

1. Introduction

Running is a steady and moderately paced running motion, and it is often considered to be a popular recreational activity and common component of many fitness routines [1]. It requires the precise coordination and synchronization of various muscles throughout the body [2]. The complexity and high redundancy of the human musculoskeletal system execute a wide variety of daily tasks [3]. Understanding the mechanics of motor control during movement can enhance our ability to optimize performance, prevent injuries, and develop effective rehabilitation strategies.

Muscle synergies are characterized by coordinated recruitment of specific muscle groups with precise timing and amplitude synchronization to achieve purposeful movements [4]. This conceptual framework provides a method to deal with the complexity of motor control and has demonstrated practical applications in various fields [5,6,7]. Although muscle synergies have shown great potential in various fields, some long-standing issues in the methodology of synergy analyses should be considered. No standardized criteriaregarding the selection and number of muscles have been included in studies [8], and the processing of EMG signals [9] makes a direct comparison and interpretation of the literature so far difficult.

Previous studies have typically focused on measuring EMGs within a limited range of eight to nineteen muscles in the arm [10,11], or eight to sixteen muscles during walking [12,13]. However, relying on a subset of muscles for a muscle synergy analysis may not fully capture the recruitment and activation patterns of all the relevant muscles [8,14]. Notably, the number and selection of these muscles, especially deep muscles, are essential for the results of a muscle synergy analysis [15]. Deep muscles play acrucial role in running by providing stability, control, and efficient force transmission throughout the running motion. Deep muscles are particularly important for accurate muscle synergy analyses, and understanding the importance of deep muscles in running has practical implications for training and injury prevention.

In summary, the purpose of this study was to determine the influence of these deep muscles on a muscle synergy analysis during running. This investigation allowed us to gain insights into the influence of muscle selection on the analysis of muscle synergies and provided a basis for a more comprehensive understanding the complexity of motor control. This analysis would shed light on muscle synergy and its contribution to running, providing valuable insights into the underlying mechanisms, optimizing training strategies, and running performance.

2. Materials and Methods

We utilized publicly available datasets [16] to analyze the running activity of one subject. The dataset consisted of ten channels of sEMG signals and kinematic and kinetic data while running at a speed of 2 m/s.

Table 1. The acronym of muscles during the study, including measured sEMG serial numbers 1 to 10, along with the overall activations of 23 simulated muscles 1 to 23.

Num	Muscle	Acronym	Num	Muscle	Acronym	Num	Muscle	Acronym
1	Biceps femoris long head	Bflh	9	Vastus lateralis	Vaslat	17	Gluteus maximus2	Glutmax2
2	Gastrocnemius lateralis	Gaslat	10	Vastus medialis	Vasmed	18	Gluteus maximus3	Glutmax3
3	Gastrocnemius medialis	Gasmed	11	Gluteus medius2	Glutmed2	19	Psoas major	Psoas
4	Gluteus maximus1	Glutmax1	12	Gluteus medius3	Glutmed3	20	Vastus intermedius	Vasint
5	Gluteus medius1	Glutmed1	13	Semimembranosus	Semimem	21	Tibialis posterior	Tibpost
6	Rectus femoris	Recfem	14	Semitendinosus	Semiten	22	Peroneus longus	Perlong
7	Soleus	Soleus	15	Biceps femoris short head	Bifemsh	23	Peroneus tertius	Pertert
8	Tibialis anterior	Tibant	16	Pectineus	Pect

presents the measurement of the sEMG signals and the simulated activations of twenty-three muscles, including the deep muscles. To obtain simulated muscle activations, an open-source platform OpenSim [17] simulation was applied to calculate the activation patterns of the various muscles involved in lower limb movements, specifically the hip, knee, and ankle. In addition to the 10 sEMG measurements, we incorporated thirteen deep muscles into the synergy analysis. By integrating surface EMG measurements and simulated deep muscle activations, we calculated and examined the synergy among these 23 muscles. This approach enabled us to capture the complete information on the muscle activations during running. While data from one subject were used in this study, they provide valuable information regarding muscle synergies during running at a speed of 2 m/s.

To obtain the sEMG envelope signals, which can reflect the level of lower extremity muscle contraction,, the sEMG signals of a single-gait cycle were digitally filtered at 50 Hz with a zero-phase 4th-order Butterworth filter, full-wave rectified, and low-pass filtered at 10 Hz with a zero-phase 4th-order Butterworth filter. The envelopes were normalized to the maximum value of each muscle.

Muscle synergies were applied via a non-negative matrix factorization (NMF) algorithm to extract from the sEMGs and simulated muscle-activated signals [6,10]. NMF enabled the decomposition of the sEMG signals into muscle synergy vectors (W) and temporal activity patterns (C) through a linear combination, as illustrated by the following equation:

$$E_{m\times n}=W_{m\times s}\times C_{s\times n}+e,\hspace{1em}s\le m$$

(1)

Here, E is an m×n representing the pre-processed EMG matrix recorded from m muscles. W is an m×s synergy matrix with reduced dimensions. The variable s represents the number of synergies to be identified from the data. C is an m×n a matrix with s activation coefficients, representing the temporal activity patterns corresponding to each synergy. Finally, e denotes the residual matrix, which accounts for any remaining noise.

To reconstruct the sEMG signals and determine the minimum number of required muscle synergies, the variability accounted for (VAF) criterion was employed [14]. A fixed threshold of VAF > 92% was used as the basis on the VAF curve to determine the number of muscle synergies that needed to be extracted.

3. Results

Our study identified four synergies that accounted for 92% of the total variance in the muscle activity when ten muscles were analyzed. These muscle synergies, represented by synergy vectors W1–W4 (Figure 1), are essential for the coordination of muscle actions during running. The activation of the Gasmed1 and Tibant muscles primarily characterizes synergy vector 1. These muscles contribute to hip abduction and dorsiflexion. The synergy activation of these muscles is essential to stabilizing the pelvis and hip joints during running, providing significant leg propulsion, and ensuring balance and postural stability throughout running. Synergy Vector 2 is mainly characterized by the synchronized activation of the Gasmax1, Gasmed1, Soleus, Vaslat, and Vasmed muscles, contributing to hip abduction, hip extension, and knee extension actions. This synergy helps to maintain stability, generate power, and prevent excessive joint movement or rotation during running. Synergy vector 3 is characterized by the activation of the hamstrings and soleus muscles. The coordinated activation of these muscles contributes to specific aspects of running mechanics. Synergy vector 4 primarily consists of the activity of Bflth and Glamax, which contribute to knee flexion and hip extension.

Then, we calculated from all 23 muscles, including 13 deep muscles, to identify the muscle synergies. As shown in Figure 2, the results showed that these five synergies accounted for 92% of the variability in muscle activation during running. Importantly, when comparing the results obtained with the 23 muscles to those obtained with 10 muscles, we observed differences in the grouping and relative weights of the muscles. The simulation study further demonstrated the contribution of the deep muscles to muscle synergy analysis, leading to the identification of an additional synergy vector. Notably, deep dominant muscles in the synergy vectors highlighted the inclusion of muscles not considered in the experimental analysis, such as the Semimem and Semiten muscles. These findings underscore the importance of considering a comprehensive set of muscles, including deep muscles, to capture the full complexity of muscle synergies in the context of the activity being studied.

4. Conclusions

In this study, we examined the impact of the selection of muscles on a muscle synergy analysis. Specifically, we compared the results obtained from 23 simulated muscles with the traditional approach of using recorded sEMG signals. Additional synergies were observed when more deep muscles were involved in the analysis. This suggests that deeper muscle activation patterns and coordination play a significant role in the overall muscle synergy dynamics during a task. These results highlight the importance of considering the contribution of deep muscles when studying muscle synergies and provide further insights into the complexity of motor control during the examined activity. Neglecting specific muscles may lead to an underestimation of the complexity of the muscle coordination, which could have implications for both laboratory research and clinical applications. It is important to note that these preliminary findings warrant further investigation, using high-density experimental data for validation.