Surface electromyography (sEMG) is one of the most commonly used measures of muscle activation. The methods of biosignal processing and analysis, such as: intermuscular coherence (IMC), Continuous Wavelet Transform (CWT), Cross-correlation offer increasingly accurate ways to monitor individual components of the recorded signal and provide insightful analysis of working muscles during motor activities (Kenville et al., 2020). While EMG allows the characterization of working muscles, intermuscular coherence and synchronization analysis of the EMG signal can explain the phenomenon of neuromuscular conduction during motor activity (d'Avella et al., 2003; Mohr et al., 2018). EMG detects the electrical potentials generated by muscle cells when activated. The motor unit action potential is a compound potential representing the sum of the individual action potentials. Motor units must synchronize their connections in order to tone the entire muscle smoothly, which requires common presynaptic stimulation of motor neurons. A popular method for testing motor control is intermuscular coherence (IMC). Using IMC, it is possible to non-invasively study the common synaptic signal for pools of motor neurons in muscles (Gross et al., 2002; Dideriksen et al., 2018). Any recorded signal, including EMG, can be viewed as the sum of determined and random components. The former carry information about the recorded signal. The latter contain information about noise and artifacts during signal recording. To separate the signal, a wavelet transform is used, which consists of selecting the wavelet that is most similar to the recorded signal, called the mother wavelet. This results in the separation of parts of the signal that are similar to the selected pattern. The operation of a wavelet is similar to that of a low-pass filter. The wavelet transform involves the decomposition of the signal into the approximation and detail coefficients. The approximation represents the component that carries spectral diagnostic information. The result is a frequency spectrum map (Ngui et al., 2013; Ghofrani Jahromi et al., 2017).

In the present study, the Continuous Wavelet Transform (CWT) method was used. The CWT effectively identifies linear coupling between EMG signals during different movement sequences. This method provides information about the level and strength of coupling in the form of time-frequency maps. On this basis, the associated characteristics of the signals in specific frequency bands can be visualized (Garg et al., 2013; Rocha et al., 2022). One of the most reliable techniques for assessing the level of motor unit synchronization in human muscles is the cross-correlation of the discharge times of two simultaneously active motor units. Cross-correlation has become the most widely used method for determining the interdependence of human motor unit discharges (Semmler, 2002). The frequency domain equivalent of cross-correlation analysis has been used in human motor unit studies to detect periodic triggering of common synaptic input signals to motor neurons. Studies have shown that coherence can be detected between pairs of motor units in the hand muscle in frequency bands of 1–12 Hz and 16–32 Hz during voluntary isometric abduction of the index finger. The observed significant coherence between pairs of motor units at these frequencies implies a common periodicity of the presynaptic signal (Farmer et al., 1993; Raikova et al., 2021). Commonly studied frequency bands include the alpha (8–15 Hz), beta (15–30 Hz), and low gamma (30–60 Hz) bands, as coherence within these bands is thought to originate from different neural sources (Kim et al., 2001). These bands are generally used to represent common neuronal inputs associated with corticospinal pathways (Weersink et al., 2021). The coherence of the alpha band originates subcortically, and the alpha band is thought to specifically reflect the involvement of the reticulospinal tract (Laine et al., 2021).

The coherence of the alpha band is related to the degree to which the overall level of muscle activity is correlated over time (Grosse and Brown, 2003). In contrast, the beta band is associated with corticospinal conduction and reflects the synchronization between the motor cortex and the working muscles. Intermuscular coherence between flexor and extensor muscle pairs is mainly observed in the beta band (Hu et al., 2018). Intermuscular coherence in the gamma band is due to cortical signals and is thought to be of functional importance in efferent conduction (Farmer et al., 1998).

The aim of the study was to investigate neuromuscular conduction in wheelchair fencers using the EMG signal from their upper limb muscles. For this purpose, the bands (1) 2–16 Hz, (2) 17–30 Hz, and (3) 31–60 Hz were selected to assess the activation of opposite muscle groups of the upper extremities: the biceps/triceps for arm movement and the flexor carpi radialis/extensor carpi radialis for forearm movement.

The purpose of comparing category A and category B fencers was to explore how varying levels of muscle activation (resulting from differences in impairments) affect performance in fencing. Category A athletes, who generally experience less severe impairments, are able to activate muscles involved in fencing more fully and may demonstrate muscle patterns more typical of able-bodied athletes. On the other hand, category B athletes often have limited or altered muscle activation due to neurological or musculoskeletal conditions, which may lead to adaptive strategies in how they control their weapons and execute movements. The procedure used resulted in a normalized measure of correlation in the frequency domain, ranging from 0 to 1. Coherence was calculated for frequencies up to 60 Hz. The analysis illustrates the individual time-dependent coherence between two signals at different frequencies during the muscle work cycle. A statistically significant difference in alpha coherence (2–16 Hz) was found in the forearm muscle group (Table 3), and its peaks were observed during periods of the fencing action cycle (Figure 2). Some differences were also found in the gamma coherence of the arm muscle signals (Table 2). Conversely, no significant activity was observed in the beta frequency band. Motor unit synchronization in the beta (15–30 Hz) and low gamma (30–60 Hz) bands was mainly driven by the primary motor cortex and less affected by the premotor cortices and supplementary motor area (Marsden et al., 2001; Chen et al., 2018; Obata et al., 2014). The results confirm the line with the other studies in which alpha reflects the reticulospinal tract, responsible for stimulation of the distal muscles of the wrist and hand, while gamma is derived from cortical signals and is related to efferent conduction and reflects corticomuscular coupling. The reticulospinal tract modulates muscle tension and reflex activity. It also causes reciprocal inhibition of antagonistic muscles. According to current knowledge, muscular synchronization can be altered by the loss or reduction of correlated neuronal inputs from the central nervous system; in such cases, motor units typically operate at a frequency of 5 to 12 Hz, indicating that there is a correlation between the oscillation generated in the CNS and the firing rate of the motor unit (Person and Kudina, 1972).

Irregularities in cortical control are responsible for the loss of alpha conduction in individuals with muscle paralysis and hemiparesis. The existence of coherence between low-gamma muscles can be explained by the need to quickly integrate information under dynamic conditions and generate appropriate motor commands. It reflects conduction between cortical activity and working muscles. The corticoreticulospinal tract (CReST) is a major descending motor pathway in many animals, but little is known about its innervation patterns in proximal and distal upper extremity muscles in humans (Taga et al., 2021). Recent studies conducted in monkeys have shown that the reticulospinal projections innervate the intrinsic muscles of the hand, with the strength of the wrist and hand muscles rivaling that of the more proximal muscles. The gamma, on the other hand, is associated with efferent conduction. In recent years, muscle synergy analysis has been widely used to determine how muscles are coordinated during complex motor activities. Results have shown that children with spastic cerebral palsy activate fewer synergies, for example in gait patterns, than children without neurological deficits (damage) (Goudriaan et al., 2022). Kisiel-Sajewicz et al. (2011), reported that people with limbic paralysis showed weaker intermuscular coupling in the low-frequency range (0–11 Hz), but these were stroke patients with movement problems. In contrast, Fisher et al. found that IMC in the beta band (15–30 Hz) can be used as a biomarker to assess upper limb motor dysfunction in stroke patients (Fisher et al., 2012). In a study of athletes with disabilities, IMC was obtained in low frequency bands. Despite the above findings, one of the main limitations of muscle synergy analysis is that it can only reflect the spatiotemporal anatomical coordination of multiple muscles and assess the activation state of overall muscle activity (Liu et al., 2022), but it does not reveal the neurophysiological mechanisms underlying the formation of muscle synergies (d'Avella et al., 2003). Coherence analysis is an effective method for assessing neural synchronization between muscles or brain areas and is usually based on cross-correlation between signals in the frequency domain (Li et al., 2016).

The human upper limb is capable of performing a variety of functional tasks due to its biomechanical structure and abundant motor and sensory innervation. Previous studies of the human arm have shown altered kinematics of the hemiparetic arm and a reduced ability to generate the required torques in upper limb joint interactions (Cirstea et al., 2003). Abnormal neurological coupling as part of flexion synergy between the elbow and shoulder muscles has also been described after stroke (Ellis et al., 2017). These synergies have also been shown to extend to the finger flexors. Indeed, the number of distinct activation patterns appears to be reduced after nervous system damage (McPherson and Dewald, 2019). However, most of the studies were based on the evaluation of patients with motor limitations. The present study is based on the performance of highly trained Paralympic athletes. It is worth noting that neural synchronization has been recognized as a mechanism that integrates the distributed sensory and motor systems involved in the coordination of complex movements. Muscle synergy represents the anatomical coordination of muscle activation over time during the performance of a specific motor task. This means that oscillatory coupling and synchronous neuronal discharges play an important role in the process of human movement control (d'Avella et al., 2003). It was Bernstein who first proposed that the central nervous system would be able to jointly activate a group of muscles to overcome the “many degrees of freedom” problem in motor control (Bernstein, 1966). D'Avella et al. (2003), proposed that the activation of interacting muscle groups is coordinated by descending correlated neural excitations. Studies from the perspective of neural oscillations conveying motor control information have suggested that IMC may reflect correlated intermuscular neural inputs in healthy subjects (Fisher et al., 2012; De Marchis et al., 2015; Hu et al., 2018). Although existing literatures cover functionalities of muscle network or synergy patterns separately, little evidence shows their collective mechanism. Luo et al. (2024) deciphered the mechanism of synergy patterns on muscle network among lower-limb muscles. Individuals exhibiting notable deficiencies in their synergy patterns should initially engage in targeted exercises to reconstruct patterns within the local muscle network before extending the approach to the global network to fully reinstate motor abilities.

Increased neural coupling between distal and proximal muscles can affect the basic modular structures of the spinal cord that control arm and hand muscles during functional activities. Manipulating objects, a common part of performing a task, requires further coordination between the joints as the arm must position, orient, and stabilize the hand while the fingers interact with an external object. Separate neural pathways for arm and hand control have been proposed, and the extent of communication between these pathways remains unclear (Phan et al., 2022). The results of the present study suggest that the increased level of hand muscle activation at target locations requiring arm abduction and elbow extension can be explained by increased neural coupling between the forearm (flexor/extensor carpi radialis) and arm muscles (biceps/triceps) (Li et al., 2016). In the synchronization of the arm muscles, for category A wheelchair fencers (Figure 1A), there was synchronization between the activities for the 2 Hz frequency in the time intervals from 1 to 2 s and from 3 to 4 s, as well as for the 4 Hz frequency in the time intervals from 0 to 0.5 s and from 3 to 3.5 s. On the other hand, for the forearm muscles of category A wheelchair fencers (Figure 2A), coherence occurred in the range from 2 to 16 Hz, in the form of short-term discharges (with a duration of several hundred ms) at 2.5, 5 s in the 3 Hz band. There was also strong synchronization at 6–8 Hz in the form of brief moments of synchronization of wavelet components. For the category B wheelchair fencers (Figure 1B), synchronization of the arm muscles occurred in the 2 to 8 Hz range at 3 to 4 s and 6 to 7 s. This is consistent with a time course where muscle activity events occur simultaneously and are synchronized in the low frequency range. A different coherence distribution is observed for the forearm muscles of category B wheelchair fencers (Figure 2B), where strong coherence was found in the 0.5 to 4 Hz range over the entire analyzed time interval. The observed coherence over the whole analyzed interval within the forearm muscles in the group of category B fencers reflects the effect of integration with the object, i.e., the continuous holding of the weapon by the fencer. Such an effect is not observed in category A fencers, in whom distinct periods of rest and stimulation can be observed. In addition, category B wheelchair fencers show a high consistency and synchronization of EMG waveforms in both channels mapped for the forearm muscles.

The intermuscular coherence analysis revealed that homologous muscle functions are induced by common oscillatory inputs that include alpha, beta, and gamma frequencies with different synchronization patterns during different movement periods. The results of the study confirm the existence of different oscillations in the alpha and gamma frequency bands depending on motor performance and the degree of neuromuscular conduction impairment (category A and B wheelchair fencers), which explains the theory of alpha band excitation, characteristic of subcortical conduction, that reticulospinal pathways modulate muscle tension and reflex activity. They also cause reciprocal inhibition of antagonistic muscles. This is especially true in precision movements performed with forearm muscles. Unlike the gamma band, which reflects cortical conduction, corticospinal pathways carry muscle control impulses from the cortex through the spinal cord to spinal nerves and muscles, providing stability especially during fast and precise motor actions.