A Comprehensive Analysis of Multilayer Community Detection Algorithms for Application to EEG-Based Brain Networks

Multilayer network analysis in patients with juvenile myoclonic epilepsy

INTRODUCTION

We conducted a multilayer network analysis in patients with juvenile myoclonic epilepsy (JME) and healthy controls, to investigate the gray matter layer using a morphometric similarity network and analyze the white matter layer using structural connectivity.

METHODS

We enrolled 42 patients with newly diagnosed JME and 53 healthy controls. Brain magnetic resonance imaging (MRI) using a three-tesla MRI scanner, including T1-weighted imaging and diffusion tensor imaging (DTI) were performed. We created a gray matter layer matrix with a morphometric similarity network using T1-weighted imaging, and a white matter layer matrix with structural connectivity using the DTI. Subsequently, we performed a multilayer network analysis by applying graph theory.

RESULTS

There were significant differences in network at the global level in the multilayer network analysis between the groups. The average multiplex participation of patients with JME was lower than that of healthy controls (0.858 vs. 0.878, p = 0.007). In addition, several regions showed significant differences in multiplex participation at the nodal level in the multilayer network analysis. Multiplex participation in the right entorhinal cortex was lower, whereas multiplex participation in the right supramarginal gyrus was higher at the nodal level in the multilayer network analysis of patients with JME compared to healthy controls.

CONCLUSION

We demonstrated differences in network at the global and nodal levels in the multilayer network analysis between patients with JME and healthy controls. These features may be associated with the pathophysiology of JME and could help us understand the complex brain network in patients with JME.

Alteration of multilayer network perspective on gray and white matter connectivity in obstructive sleep apnea

PURPOSE

The objective of this research was to examine changes in the neural networks of both gray and white matter in individuals with obstructive sleep apnea (OSA) in comparison to those without the condition, employing a comprehensive multilayer network analysis.

METHODS

Patients meeting the criteria for OSA were recruited through polysomnography, while a control group of healthy individuals matched for age and sex was also assembled. Utilizing T1-weighted imaging, a morphometric similarity network was crafted to represent gray matter, while diffusion tensor imaging provided structural connectivity for constructing a white matter network. A multilayer network analysis was then performed, employing graph theory methodologies.

RESULTS

We included 40 individuals diagnosed with OSA and 40 healthy participants in our study. Analysis revealed significant differences in various global network metrics between the two groups. Specifically, patients with OSA exhibited higher average degree overlap and average multilayer clustering coefficient (28.081 vs. 23.407, p < 0.001; 0.459 vs. 0.412, p = 0.004), but lower multilayer modularity (0.150 vs. 0.175, p = 0.001) compared to healthy controls. However, no significant differences were observed in average multiplex participation, average overlapping strength, or average weighted multiplex participation between the patients with OSA and healthy controls. Moreover, several brain regions displayed notable differences in degree overlap at the nodal level between patients with OSA and healthy controls.

CONCLUSION

Remarkable alterations in the multilayer network, indicating shifts in both gray and white matter, were detected in patients with OSA in contrast to their healthy counterparts. Further examination at the nodal level unveiled notable changes in regions associated with cognition, underscoring the effectiveness of multilayer network analysis in exploring interactions across brain layers.

Multilayer network-based channel selection for motor imagery brain-computer interface

Objective. The number of electrode channels in a motor imagery-based brain-computer interface (MI-BCI) system influences not only its decoding performance, but also its convenience for use in applications. Although many channel selection methods have been proposed in the literature, they are usually based on the univariate features of a single channel. This leads to a loss of the interaction between channels and the exchange of information between networks operating at different frequency bands.Approach. We integrate brain networks containing four frequency bands into a multilayer network framework and propose a multilayer network-based channel selection (MNCS) method for MI-BCI systems. A graph learning-based method is used to estimate the multilayer network from electroencephalogram (EEG) data that are filtered by multiple frequency bands. The multilayer participation coefficient of the multilayer network is then computed to select EEG channels that do not contain redundant information. Furthermore, the common spatial pattern (CSP) method is used to extract effective features. Finally, a support vector machine classifier with a linear kernel is trained to accurately identify MI tasks.Main results. We used three publicly available datasets from the BCI Competition containing data on 12 healthy subjects and one dataset containing data on 15 stroke patients to validate the effectiveness of our proposed method. The results showed that the proposed MNCS method outperforms all channels (85.8% vs. 93.1%, 84.4% vs. 89.0%, 71.7% vs. 79.4%, and 72.7% vs. 84.0%). Moreover, it achieved significantly higher decoding accuracies on MI-BCI systems than state-of-the-art methods (pairedt-tests,p< 0.05).Significance. The experimental results showed that the proposed MNCS method can select appropriate channels to improve the decoding performance as well as the convenience of the application of MI-BCI systems.

Alterations in the multilayer network in patients with rapid eye movement sleep behaviour disorder

This study aimed to reveal the pathophysiology of isolated rapid eye movement sleep behaviour disorder (RBD) in patients using multilayer network analysis. Participants eligible for isolated RBD were included and verified via polysomnography. Both iRBD patients and healthy controls underwent brain MRI, including T1-weighted imaging and diffusion tensor imaging. Grey matter matrix was derived from T1-weighted images using a morphometric similarity network. White matter matrix was formed from diffusion tensor imaging-based structural connectivity. Multilayer network analysis of grey and white matter was performed using graph theory. We studied 29 isolated RBD patients and 30 healthy controls. Patients exhibited a higher average overlap degree (27.921 vs. 23.734, p = 0.002) and average multilayer clustering coefficient (0.474 vs. 0.413, p = 0.002) compared with controls. Additionally, several regions showed significant differences in the degree of overlap and multilayer clustering coefficient between patients with isolated RBD and healthy controls at the nodal level. The degree of overlap in the left medial orbitofrontal, left posterior cingulate, and right paracentral nodes and the multilayer clustering coefficients in the left lateral occipital, left rostral middle frontal, right fusiform, right inferior posterior parietal, and right parahippocampal nodes were higher in patients with isolated RBD than in healthy controls. We found alterations in the multilayer network at the global and nodal levels in patients with isolated RBD, and these changes may be associated with the pathophysiology of isolated RBD. Multilayer network analysis can be used widely to explore the mechanisms underlying various neurological disorders.

Relation of connectome topology to brain volume across 103 mammalian species

The brain connectome is an embedded network of anatomically interconnected brain regions, and the study of its topological organization in mammals has become of paramount importance due to its role in scaffolding brain function and behavior. Unlike many other observable networks, brain connections incur material and energetic cost, and their length and density are volumetrically constrained by the skull. Thus, an open question is how differences in brain volume impact connectome topology. We address this issue using the MaMI database, a diverse set of mammalian connectomes reconstructed from 201 animals, covering 103 species and 12 taxonomy orders, whose brain size varies over more than 4 orders of magnitude. Our analyses focus on relationships between volume and modular organization. After having identified modules through a multiresolution approach, we observed how connectivity features relate to the modular structure and how these relations vary across brain volume. We found that as the brain volume increases, modules become more spatially compact and dense, comprising more costly connections. Furthermore, we investigated how spatial embedding shapes network communication, finding that as brain volume increases, nodes' distance progressively impacts communication efficiency. We identified modes of variation in network communication policies, as smaller and bigger brains show higher efficiency in routing- and diffusion-based signaling, respectively. Finally, bridging network modularity and communication, we found that in larger brains, modular structure imposes stronger constraints on network signaling. Altogether, our results show that brain volume is systematically related to mammalian connectome topology and that spatial embedding imposes tighter restrictions on larger brains.

Multilayer network changes in patients with migraine

INTRODUCTION

To investigate changes in the multilayer network in patients with migraine compared to healthy controls.

METHODS

This study enrolled 82 patients with newly diagnosed migraine without aura and 53 healthy controls. Brain magnetic resonance imaging (MRI) was conducted using a 3-tesla MRI scanner, including three-dimensional T1-weighted and diffusion tensor imaging (DTI). A gray matter layer matrix was created with a morphometric similarity network using T1-weighted imaging and the FreeSurfer program. A white matter layer matrix was also created with structural connectivity using the DTI studio (DSI) program. A multilayer network analysis was then performed by applying graph theory using the BRAPH program.

RESULTS

Significant changes were observed in the multilayer network at the global level in patients with migraines compared to the healthy controls. The multilayer modularity (0.177 vs. 0.160, p = .0005) and average multiplex participation (0.934 vs. 0.924, p = .002) were higher in patients with migraines than in the healthy controls. In contrast, the average multilayer clustering coefficient (0.406 vs. 0.461, p = .0005), average overlapping strength (56.061 vs. 61.676, p = .0005), and average weighted multiplex participation (0.847 vs. 0.878, p = .0005) were lower in patients with migraine than in the healthy controls. In addition, several regions showed significant changes in the multilayer network at the nodal level, including multiplex participation, multilayer clustering coefficients, overlapping strengths, and weighted multiplex participation.

CONCLUSION

This study demonstrated significant changes in the multilayer network in patients with migraines compared to healthy controls. This could aid an understanding of the complex brain network in patients with migraine and may be associated with the pathophysiology of migraines. Patients with migraine show multilayer network changes in widespreading brain regions compared to healthy controls, and specific brain areas seem to play a hub role for pathophysiology of the migraine.

Involvement of the default mode network in patients with transient global amnesia: multilayer network

INTRODUCTION

We aimed to investigate the alterations in the multilayer network in patients with transient global amnesia (TGA).

METHODS

We enrolled 124 patients with TGA and 80 healthy controls. Both patients with TGA and healthy controls underwent a three-teslar brain magnetic resonance imaging (MRI). A gray matter layer matrix was created using a morphometric similarity network derived from the T1-weighted imaging, and a white matter layer matrix was constructed using structural connectivity based on the diffusion tensor imaging. A multilayer network analysis was performed by applying graph theoretical analysis.

RESULTS

There were no significant differences in global network measures between the groups. However, several regions, related to the default mode network, showed significant differences in nodal network measures between the groups. Multi-richness in the left pars opercularis, multi-rich-club degree in the right posterior cingulate gyrus, and weighted multiplex participation in the right posterior cingulate gyrus were higher in patients with TGA compared with healthy controls (15.47 vs. 12.26, p = 0.0005; 41.68 vs. 37.16, p = 0.0005; 0.90 vs. 0.80, p = 0.0005; respectively). The multiplex core-periphery in the left precuneus was higher (0.96 vs. 0.84, p = 0.0005), whereas that in the transverse temporal gyrus was lower in patients with TGA compared with healthy controls (0.00 vs. 0.02, p = 0.0005).

CONCLUSION

We newly find the alterations in the multilayer network in patients with TGA compared with healthy controls, which shows the involvement of the default mode network. These changes may be related to the pathophysiology of TGA.

Community detection in multi-frequency EEG networks

Functional connectivity networks of the human brain are commonly studied using tools from complex network theory. Existing methods focus on functional connectivity within a single frequency band. However, it is well-known that higher order brain functions rely on the integration of information across oscillations at different frequencies. Therefore, there is a need to study these cross-frequency interactions. In this paper, we use multilayer networks to model functional connectivity across multiple frequencies, where each layer corresponds to a different frequency band. We then introduce the multilayer modularity metric to develop a multilayer community detection algorithm. The proposed approach is applied to electroencephalogram (EEG) data collected during a study of error monitoring in the human brain. The differences between the community structures within and across different frequency bands for two response types, i.e. error and correct, are studied. The results indicate that following an error response, the brain organizes itself to form communities across frequencies, in particular between theta and gamma bands while a similar cross-frequency community formation is not observed following the correct response.

Community detection in brain connectomes with hybrid quantum computing

Recent advancements in network neuroscience are pointing in the direction of considering the brain as a small-world system with an efficient integration-segregation balance that facilitates different cognitive tasks and functions. In this context, community detection is a pivotal issue in computational neuroscience. In this paper we explored community detection within brain connectomes using the power of quantum annealers, and in particular the Leap's Hybrid Solver in D-Wave. By reframing the modularity optimization problem into a Discrete Quadratic Model, we show that quantum annealers achieved higher modularity indices compared to the Louvain Community Detection Algorithm without the need to overcomplicate the mathematical formulation. We also found that the number of communities detected in brain connectomes slightly differed while still being biologically interpretable. These promising preliminary results, together with recent findings, strengthen the claim that quantum optimization methods might be a suitable alternative against classical approaches when dealing with community assignment in networks.

Multi-modal and multi-subject modular organization of human brain networks

The human brain is a complex network of anatomically interconnected brain areas. Spontaneous neural activity is constrained by this architecture, giving rise to patterns of statistical dependencies between the activity of remote neural elements. The non-trivial relationship between structural and functional connectivity poses many unsolved challenges about cognition, disease, development, learning and aging. While numerous studies have focused on statistical relationships between edge weights in anatomical and functional networks, less is known about dependencies between their modules and communities. In this work, we investigate and characterize the relationship between anatomical and functional modular organization of the human brain, developing a novel multi-layer framework that expands the classical concept of multi-layer modularity. By simultaneously mapping anatomical and functional networks estimated from different subjects into communities, this approach allows us to carry out a multi-subject and multi-modal analysis of the brain's modular organization. Here, we investigate the relationship between anatomical and functional modules during resting state, finding unique and shared structures. The proposed framework constitutes a methodological advance in the context of multi-layer network analysis and paves the way to further investigate the relationship between structural and functional network organization in clinical cohorts, during cognitively demanding tasks, and in developmental or lifespan studies.

A novel explainable COVID-19 diagnosis method by integration of feature selection with random forest

Several Artificial Intelligence-based models have been developed for COVID-19 disease diagnosis. In spite of the promise of artificial intelligence, there are very few models which bridge the gap between traditional human-centered diagnosis and the potential future of machine-centered disease diagnosis. Under the concept of human-computer interaction design, this study proposes a new explainable artificial intelligence method that exploits graph analysis for feature visualization and optimization for the purpose of COVID-19 diagnosis from blood test samples. In this developed model, an explainable decision forest classifier is employed to COVID-19 classification based on routinely available patient blood test data. The approach enables the clinician to use the decision tree and feature visualization to guide the explainability and interpretability of the prediction model. By utilizing this novel feature selection phase, the proposed diagnosis model will not only improve diagnosis accuracy but decrease the execution time as well.

Multilayer Network Approach in EEG Motor Imagery with an Adaptive Threshold

The brain has been understood as an interconnected neural network generally modeled as a graph to outline the functional topology and dynamics of brain processes. Classic graph modeling is based on single-layer models that constrain the traits conveyed to trace brain topologies. Multilayer modeling, in contrast, makes it possible to build whole-brain models by integrating features of various kinds. The aim of this work was to analyze EEG dynamics studies while gathering motor imagery data through single-layer and multilayer network modeling. The motor imagery database used consists of 18 EEG recordings of four motor imagery tasks: left hand, right hand, feet, and tongue. Brain connectivity was estimated by calculating the coherence adjacency matrices from each electrophysiological band (δ, θ, α and β) from brain areas and then embedding them by considering each band as a single-layer graph and a layer of the multilayer brain models. Constructing a reliable multilayer network topology requires a threshold that distinguishes effective connections from spurious ones. For this reason, two thresholds were implemented, the classic fixed (average) one and Otsu’s version. The latter is a new proposal for an adaptive threshold that offers reliable insight into brain topology and dynamics. Findings from the brain network models suggest that frontal and parietal brain regions are involved in motor imagery tasks.

Modularity maximization as a flexible and generic framework for brain network exploratory analysis

The modular structure of brain networks supports specialized information processing, complex dynamics, and cost-efficient spatial embedding. Inter-individual variation in modular structure has been linked to differences in performance, disease, and development. There exist many data-driven methods for detecting and comparing modular structure, the most popular of which is modularity maximization. Although modularity maximization is a general framework that can be modified and reparamaterized to address domain-specific research questions, its application to neuroscientific datasets has, thus far, been narrow. Here, we highlight several strategies in which the "out-of-the-box" version of modularity maximization can be extended to address questions specific to neuroscience. First, we present approaches for detecting "space-independent" modules and for applying modularity maximization to signed matrices. Next, we show that the modularity maximization frame is well-suited for detecting task- and condition-specific modules. Finally, we highlight the role of multi-layer models in detecting and tracking modules across time, tasks, subjects, and modalities. In summary, modularity maximization is a flexible and general framework that can be adapted to detect modular structure resulting from a wide range of hypotheses. This article highlights multiple frontiers for future research and applications.

Multi-layer analysis of multi-frequency brain networks as a new tool to study EEG topological organization

Oscillatory activity rising from the interaction among neurons is widely observed in the brain at different scales and is thought to encode distinctive properties of the neural processing. Classical investigations of neuroelectrical activity and connectivity usually focus on specific frequency bands, considered as separate aspects of brain functioning. However, this might not paint the whole picture, preventing to see the brain activity as a whole, as the result of an integrated process. This study aims to provide a new framework for the analysis of the functional interaction between brain regions across frequencies and different subjects. We ground our work on the latest advances in graph theory, exploiting multi-layer community detection. In our multi-layer network model, layers keep track of single frequencies, including all the information in a unique graph. Community detection is then applied by means of a multilayer formulation of modularity. As a proof-of-concept of our approach, we provide here an application to multi-frequency functional brain networks derived from resting state EEG collected in a group of healthy subjects. Our results indicate that α-band selectively characterizes an inter-individual common organization of EEG brain networks during open eyes resting state. Future applications of this new approach may include the extraction of subject-specific features able to capture selected properties of the brain processes, related to physiological or pathological conditions.