Multiple sclerosis (MS) is an autoimmune disease with disseminated cortical and white-matter lesions and devastating consequences for the patients (Dendrou et al., 2015). In the nervous system, axonal fibers provide the anatomical basis for signal transfer and communication between different regions. The damaged myelin causes communication problems between different regions of the CNS. Further processes of remyelination and structural and functional rearrangement are hallmarks of the disease (Filippi and Agosta, 2009). Focal demyelination is spread over the entire brain, primarily involving white matter, with predilection sites in the periventricular regions. In contrast, gray matter lesions are increasingly considered significant trademarks of the disease and play an essential role in the long-term functional outcome (Seewann et al., 2011; Droby et al., 2015, 2016). Due to the damaged axonal fibers and communication problems between various regions, patients with MS face severe daily life challenges. Generally, patients with MS show sensory impairments including pain and unpleasant sensations; an imbalance that can arise from cerebellar dysfunction, cognitive impairments, and slowed information processing; depression that may relate to injury of frontotemporal networks; fatigue, bladder, and bowel dysfunction; heat sensitivity; headache; and pseudo relapse. Furthermore, brain stem syndromes such as double vision and facial weakness are common in these patients. Relapsing-remitting MS (RRMS) is the most common form of MS in which relapses are discrete periods of neurologic dysfunction that commonly evolve over hours to days and then persist for days to weeks before remitting. In between relapses, patients are generally stable but still often experience fatigue and heat sensitivity (Gelfand, 2014). Previous studies have used multimodal evoked potentials as a robust monitoring tool to compare the ability of different evoked responses to investigate and predict the disease course (Schlaeger et al., 2016).

In addition to task-directed studies, another approach to investigate brain activity is to gather data during the resting state when the subjects' nervous system is not responding to a specific stimulation (Biswal et al., 1995). In previous years, some studies have focused on monitoring brain activity in patients with MS during the resting state to investigate the abnormalities using fMRI to reach a deeper understanding of the impacts of MS on the brain's function (Liu et al., 2011; Tahedl et al., 2018; Saccà et al., 2019). However, although neuroimaging methods such as MRI and fMRI have a high spatial resolution due to their data acquisition method, they are usually expensive for continuous recordings. Continuous recordings help researchers to monitor disease progression or examine the effects of different treatments on patients for an extended period. Furthermore, using fMRI in studying the brain's function can be time-consuming, which can cause various problems for patients with MS due to their physical condition, making the data acquisition process arduous. Moreover, another problem with fMRI-based methods is that they have a low temporal resolution that can lead to information loss, which is not ideal for monitoring the activity of a dynamic system like the brain.

Electroencephalography can be a decent method to monitor brain function as its recording cost is less than other neuroimaging methods. In previous years, researchers have investigated EEG resting-state to monitor the brain's cortical activity and compare the brain activity differences between healthy people and MS phenotypes in different frequency bands (Babiloni et al., 2016). The results of these studies show differences in the activity of various frequency bands between MS phenotypes and healthy cases in various cortical regions.

Mapping the active functional properties of brain networks is a crucial goal of basic and clinical neuroscience and medicine. Brain network theory has been repeatedly used to understand brain functions. This theory explains that the brain's complex functions result from the communication between different brain regions (Lang et al., 2012; Sporns, 2016). One of the powerful tools for quantifying and analyzing the characteristics of connectivity networks is graph theory. Graph theory has become one of the most useful methods for modeling brain networks as interconnected systems in recent years (Bassett and Sporns, 2017). The graph theory approach has been proven worthy in various brain network research for subjects with both normal and pathological conditions such as major depressive disorder and Alzheimer's disease (Brier et al., 2014; Hasanzadeh et al., 2020). Network science contributes powerful access to primary organizational principles of the human brain, and it has been applied with graph theory to distinguish brain structural and functional connectivity patterns in MS (Fleischer et al., 2019). Previous studies have used the graph theory approach to investigate structural connectivity in MS and introduce methods to classify patients with MS in several clinical profiles (Kocevar et al., 2016). fMRI and MEG-based neuroimaging studies have used the graph theory approach for studying functional connectivity in MS to assess impaired functional integration and characterize topological patterns of connectivity changes (Schoonheim et al., 2013; Dogonowski et al., 2014; Rocca et al., 2016).

According to previous EEG studies that investigated brain activity during resting state in MS cases, we hypothesize that EEG resting-state can show significant differences in brain functional network characteristics between MS and healthy cases. To validate this hypothesis, we constructed the brain network by calculating connectivity measures between EEG channels after preprocessing. In this study, EEG-based networks have been constructed to compare the brain network between RRMS and healthy cases. The directed transfer function (DTF) and imaginary part of coherence (iCoh) were chosen to investigate the information flow and their activity synchronization. The rest of this study has been organized into four sections: first, in the “Participants and method” section, we describe the subjects who participated in this research, then the EEG data recording and preprocessing procedure are explained. Next, we describe the method used for constructing brain networks, and then graph theory features extracted from the networks are mentioned. Next, in the “Result” section, alterations in the networks of patients with MS are represented, and the two groups' network characteristics are compared. Then, the obtained results are discussed in the “Discussion” section. Finally, this study is concluded in the “Conclusion” section.

After calculating the connectivity and construction of the brain network for each subject, we used the Shapiro-Wilk test to evaluate the normality of the distribution of results and the Wilcoxon rank-sum test to distinguish different features between the healthy and RRMS groups.

This study demonstrated a lower amplitude in the alpha peak and a higher amplitude in the gamma band for patients with RRMS than healthy subjects. In the previous studies, irregularities in individual frequency bands' activity, such as alpha, beta, and gamma, have been demonstrated. These studies have shown a lower amplitude for the alpha frequency band (Babiloni et al., 2016). The resting-state eyes-closed EEG rhythms exhibit a condition of relative relaxation and inhibition of brain mechanisms subserving the interaction with the external world. A complex pattern of parallel neural networks generates these rhythms in which the integrative activity of the neurons is characterized by a synchronous oscillatory mode across neural assemblies (Nunez, 1989; Başar, 2012). A pathological decrease in the resting-state alpha frequency band in patients with MS probably indicates a pathological desynchronization of diffuse neural networks regulating the fluctuation of cortical arousal and tonic alertness. When a person is involved in a situation that captures his awareness or implies various actions, alpha and low-frequency beta rhythms reduce amplitude, which shows a desynchronization. Here, these slower oscillations are replaced by fast oscillations at high-frequency beta and gamma (Pfurtscheller and Da Silva, 1999).

Network diameter is the shortest distance between the two most remote nodes in the network. A higher diameter implies a less-linked network. According to the DTF connectivity calculations, which designate the level of signal transformation and association between different channels, the linkage in RRMS subjects is lower than in the healthy group. These results are possibly the effect of axonal demyelination and the existence of lesions in the neural network, specifically in cortical and subcortical regions, which not only can manipulate the rate of signal transfer but also, in more severe conditions, can lead to the complete loss of communication between different regions. According to previous studies, a predominant pattern of decreased global connectivity due to acute neuroinflammation (Kocevar et al., 2016) or increased lesion load can be associated with the disease course (He et al., 2009). Several fMRI studies have shown reduced coherence in resting-state networks in MS cases (Rocca et al., 2012; Sbardella et al., 2015). Furthermore, previous studies have shown a correlation between the cortical and subcortical damage that causes cognitive impairments in MS disease with the decrease in the coherence connectivity in the alpha frequency band. In previous EEG-based studies, additional information from MRI has shown a correlation between the lesion load and the abnormal characteristics of functional brain networks related to the cognitive impairments in MS (Leocani et al., 1998, 2000).

In addition to the network linkage based on DTF, the results of this study showed significant differences in the level of global network synchronization between the two groups. According to the iCoh connectivity measure, the level of transitivity and global efficiency is generally higher in RRMS cases compared to normal subjects for the iCoh measure for the average frequency range and delta, theta, and beta frequency bands. An increase in transitivity represents a cost-efficient organization principle of the brain with increased local cliquishness (Newman and Park, 2003). Furthermore, the increase in global efficiency indicates that information transfer across the whole brain is more efficient (Latora and Marchiori, 2001). Therefore, the higher transitivity and global efficiency for iCoh indicate a higher global synchronization of the brain network in MS cases. Previous research on functional connectivity changes using MEG resting-state data indicated an increase in synchronization in the theta, lower alpha, and beta bands (Schoonheim et al., 2013).

The increased synchronization in the mentioned frequency bands and differences between DTF and iCoh measures needs further study to show what exactly it means. However, the global synchronization increase, especially in the alpha frequency band, can represent the compensation reaction of the brain in MS cases. Although in the previous studies, there have been discrepancies between the results of fMRI and MEG studies that can be explained by the fundamental differences between fMRI and MEG since fMRI cannot detect high-frequency oscillations. Previous fMRI studies have shown that functionally linked brain regions compensate for the structural damage caused by MS (Droby et al., 2016). fMRI studies have shown that when MS causes damage in a specific area, new areas are sometimes recruited to perform the tasks previously performed by the damaged area. These studies have elucidated the role of cortical reorganization in suggestive MS cases (Rocca et al., 2003; Rocca and Filippi, 2007). Furthermore, another study has shown the compensatory cortical activation during the earliest stage of MS (Audoin et al., 2003). Compensation reaction and compensation failure have been studied in MS disease for cognitive tasks during task-directed studies (cognitive compensation failure in MS). The compensation reactions of the brain have been observed in other studies, and it is not limited to MS cases (Klöppel et al., 2009; Dixon et al., 2013). A recent fMRI study showed continuous functional reorganization between specific brain regions such as fronto-cerebellar connections, which were closely related to clinical performance, indicated by increasing effective connectivity over time in MS cases, which was not noticeable in healthy cases (Fleischer et al., 2020). These results can show the role of functional reorganization in compensating for the damage and abnormalities in these cases. Moreover, results from a recent study showed a correlation between cortical plasticity and cognitive impairment in which the plasticity is reduced in cases with cognitive impairment compared to the cases with preserved cognitive function, and second, generally, the RRMS group did not show reduced cortical plasticity compared to the control group (Balloff et al., 2022). Although this study used a specific approach to investigate cortical reorganization and does not illustrate the difference between RRMS patients with normal cases, it shows that in RRMS cases without cognitive impairment, brain cortical reorganization compensates for structural damage. Considering the condition of RRMS cases in this research, the results of our study suggest that we can detect brain cortical reorganization even during the resting state while there is no specific cognitive task using surface EEG. One possible way to further investigate resting-state EEG capability for monitoring cortical reorganization and its relation to cognitive functioning is to perform a comprehensive cognitive test while recording EEG in addition to the resting-state EEG. In addition to the increase in global efficiency and transitivity for the iCoh in the alpha frequency band, the global increase in synchronization for the average of the entire frequency range might be related to damaged inhibitory neurons in MS disease (Schoonheim et al., 2013), which can lead to a higher possibility of firing rate in the network. However, it is important to remember that excitatory neurons can be damaged as well as inhibitory neurons, and further investigations are needed to explain the global increase in synchronization in RRMS subjects.

The application of EEG-based connectivity techniques in patients with MS has undoubtedly helped to improve our understanding of the mechanisms responsible for clinical disability and cognitive impairment in this condition. In this study, we used EEG resting-state signals and calculated the connectivity between EEG channels to investigate the functional brain network's global characteristics in patients with RRMS. This study's results demonstrated significant changes in brain functional network characteristics in patients with RRMS that are possibly related to damaged cortico-cortical and cortico-subcortical connections that lead to cognitive impairments in patients with MS. To have a better understanding of the relationship between the number of lesions and their location in the brain to the compensation reaction of the brain's functional network related to cognitive impairments, further studies are necessitated.

Regarding the limitations of this study, there is room for improvement in the data we recorded. First, the number of channels can be higher, which helps to have better spatial resolution and source separation. However, it needs to be mentioned that a higher number of channels increases the preparation time, making the patients uncomfortable. The second point which can be improved is that we recorded data from only the relapsing-remitting subgroup and each patient in a single session. By increasing the diversity of the patients and including other subgroups as well, and recording data from the same subjects several times, not only we can compare various stages of the disease better, but also, we can assess the effectiveness of the treatment methods with higher confidence in the future. However, it must be mentioned that other subgroups have a delicate situation due to the major development and progression of their disease, so gathering data from these groups is much more complicated and needs full permission from clinical doctors and the ethics committee. Finally, it is expected that, like any other research work, in terms of the method and pipeline we used, there is always room for improvement, which is something we look forward to in future projects related to this topic.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Iran Medical University. The patients/participants provided their written informed consent to participate in this study.

SS: analysis and interpretation of data and drafting the manuscript. MM: conception and design of study and revising the manuscript critically for important intellectual content. Both authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.