^{1}

^{2}

^{*}

^{2}

^{3}

^{4}

^{5}

^{1}

^{2}

^{3}

^{4}

^{5}

Edited by: James Danckert, University of Waterloo, Canada

Reviewed by: Gennady Knyazev, State Scientific-Research Institute of Physiology & Basic Medicine, Russia; Irene Messina, Università degli Studi di Padova, Italy

This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

Neural network-based investigations of stuttering have begun to provide a possible integrative account for the large number of brain-based anomalies associated with stuttering. Here we used resting-state EEG to investigate functional brain networks in adults who stutter (

Stuttering is a developmental disorder of speech fluency that affects 1% of all adults (

Over the last two decades, a body of neuroimaging research has amassed which suggests that stuttering likely emerges from deficiencies in the brain mechanisms that support fluent speech production (e.g.,

Studies using connectivity analyses and graph theoretical methods have demonstrated network abnormalities in stuttering during resting state (

To date, investigations of brain networks in stuttering have largely relied on functional magnetic resonance imaging (

Graph theoretical analysis (

Small-world topography (

A more recently developed approach, minimum spanning tree (

This study represents the first attempt to use quantitative

Nineteen

After recording, signals were screened for artifacts visually by an expert and then submitted to a z-score based artifact rejection algorithm implemented in the NeuroGuide software^{1}. Twenty-five artifact-free segments (each segment was between 4 and 5 s in duration) were selected and exported for

Functional connectivity between 84 regions of interests (ROIs) was obtained by ^{2}.

Non-instantaneous or lagged coherence is a methodological approach to frequency domain connectivity that removes the effects of volume conduction in

The 84 by 84 adjacency matrices were calculated separately in the theta, alpha, beta1, and beta2 bands. In the adjacency matrix, each row and each column represents a Brodmann area, and the lagged coherences between pairs of Brodmann areas are quantified at their intersections. Weighted and undirected adjacency matrices for the two groups (

Weighted and sparse (

Before 1998, graphs were classified generally as either random or regular (

In the majority of previous studies utilizing

Commonly, an adjacency brain connectivity matrix contains connectivity measures between nodes (electrodes or brain regions) that are not binary (e. g. coherence is a scalar that lies in the range 0 to 1; phase lag falls between -1 and +1, etc.). Therefore, the original adjacency brain connectivity matrix is a weighted matrix. For simplification, thresholding can be applied to transform weighted matrices to binary forms (

Where, Δ_{C}is _{lattice}-_{given} divided by C_{lattice}-C_{random} and Δ_{L}is _{given}-_{random} divided by_{lattice}-_{random}. In this equation, Lattice and random graphs have the same size (number of nodes) and the same distribution of degree (probability distribution of degrees over all nodes) within the given network (

As indicated in the above equation,

Here, random graphs with the same degree distribution and same size were made by permutation of adjacency matrices. Then

The aforementioned problems involved thresholding the connectivity matrix are overcome by transforming the original weighted matrix into a unique sparse matrix. Functional brain connectivity using the

^{3}.

Non-parametric permutation tests (

To evaluate the local corporation of cortical regions,

After

In the theta band a significant difference between groups occurred in the diameter of

In the beta2 band, a significant difference was seen in the maximum

Since there was a significant difference between groups in the maximum

Minimum spanning tree in the beta2 band. Significant differences of

Weighted and sparse

These results indicate that alterations in very fast fluctuations and synchronization of post synaptic dipole arrangements, in various brain regions involved in generating

The role of alpha activity in emotional states such as anxiety (

In the theta band,

Theta activity is closely associated with executive functions such as problem solving, planning, working memory, and also attention (

Many studies have investigated the role of motor, speech and auditory related impairments in stuttering, e.g., (

Recently it has been suggested by

Two limitations that affect this study should be considered. Firstly, as LORETA accuracy is dependent to an extent on the EEG montage density, the relatively small number of electrodes we were limited to suggest that some caution regarding interpreting absolute source localization accuracy should be exercised. However, this is always the case with EEG source analysis, and the fact that the results presented here are both physiologically plausible and in strong concordance with previous studies mitigate this concern. Furthermore, whilst there is some evidence that suggests montage density positively correlates with deep source reconstruction accuracy, a clear relationship to reconstruction of superficial sources is less clear (

Our results reinforce previous findings that DMN deficits occur in stuttering (

AG performed the main analysis, data recording, and wrote the main body of the manuscript. MA performed a part of graph analysis and he collaborated in data recording. PS managed the methodology, edited the manuscript, and also wrote a considerable part of the Section “Discussion.”

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.

We are very thankful to Professor Warren H. Meck for great comments on our presentation at Duke University.