key: cord-0297419-9bdt36ss authors: Alvand, A.; Kuruvilla-Mathew, A.; Kirk, I. J.; Roberts, R. P.; Pedersen, M.; Purdy, S. C. title: Altered brain network topology in children with Auditory Processing Disorder: a resting-state multi-echo fMRI study date: 2022-04-06 journal: nan DOI: 10.1101/2022.04.05.22273478 sha: a1974645fffcd3c543d46e78d1901cd02088e5eb doc_id: 297419 cord_uid: 9bdt36ss Children with auditory processing disorder (APD) experience hearing difficulties, particularly in the presence of competing sounds, despite having normal audiograms. There is considerable debate on whether APD symptoms originate from bottom-up (e.g., auditory sensory processing) and/or top-down processing (e.g., cognitive, language, memory). A related issue is that little is known about whether functional brain network topology is altered in APD. Therefore, we used resting-state functional magnetic resonance imaging data to investigate the functional brain network organization of 57 children from 8 to 13 years old, diagnosed with APD (n=28) and without hearing difficulties (healthy control, HC; n=29). We applied complex network analysis using graph theory to assess the whole-brain integration and segregation of functional networks and brain hub architecture. Our results showed children with APD and HC have similar global network properties (i.e., an average of all brain regions) and modular organization. Still, the APD group showed different hub architecture in default mode-ventral attention, somatomotor and frontoparietal-dorsal attention modules. At the nodal level (i.e., single-brain regions), we observed decreased participation coefficient (PC: a measure quantifying the diversity of between-network connectivity) in auditory cortical regions in APD, including bilateral superior temporal gyrus and left middle temporal gyrus. Beyond auditory regions, PC was also decreased in APD in bilateral posterior temporo-occipital cortices, left intraparietal sulcus, and right posterior insular cortex. Correlation analysis suggested a positive association between PC in the left parahippocampal gyrus and the listening-in-spatialized-noise -sentences task where APD children were engaged in auditory perception. In conclusion, our findings provide evidence of altered brain network organization in children with APD, specific to auditory networks, and shed new light on the neural systems underlying children's listening difficulties. Auditory Processing Disorder (APD) is an umbrella term for listening difficulties that result from a deficit in the neural processing of auditory stimuli or speech (Keith et al., 2019; Dawes & Bishop, 2010 ; American Speech-Language-Hearing Association (ASHA), 2005; Dillon & Cameron, 2021) . It is estimated that 5.1% of school-aged children have difficulties understanding speech in competing background noise such as the classroom despite having no hearing loss based on their pure tone audiogram (Hind et al., 2011; Purdy et al., 2018) . Keith et al. (2019) estimated that in New Zealand, where this study is conducted, the prevalence of APD in school-aged children is around 6.2% or higher. Classroom difficulties of children with APD include difficulty hearing in background noise, poor sound localization, inconsistency in answering questions, frequent requests for repetition, trouble understanding and complying with verbal instructions, and poor attention (American Academy of Audiology (AAA), de Wit et al., 2016; Chermak et al., 2002) . Some of these children also show deficits in their speech and language skills, including reading and writing (Sharma et al., 2009; Dawes & Bishop, 2010; Barker et al., 2017; Gokula et al., 2019) . In the past 40 years, children with these symptoms have been invited for further specialized testing at audiology clinics to investigate the possibility of APD (AAA, 2010) . Although there is no global consensus amongst audiologists and speech-language pathologists regarding diagnosing developmental APD , substantial efforts have been made to develop clinical guidelines for the assessment and treatment of children with APD (ASHA, 2005; Keith et al., 2019; Wilson, 2018; Iliadou et al., 2018) . Complex auditory processing happens at all levels within the auditory system; the ability to localize, discriminate, recognize auditory patterns, or discriminate temporal sound features could be impaired (Wilson, 2018; ASHA, 1996) . This impairment in auditory perception is characterized as caused by deficiencies or developmental differences in the central auditory nervous system (CANS), through which auditory signals are transmitted via the cochlear nerves to the auditory cortices (British Society of Audiology (BSA), 2011; AAA, 2010; ASHA, 1996) . Neural encoding of auditory signals is associated with complex, parallel and serial processing within auditory regions in CANS and other processing in the higher-order cortical regions (AAA, 2010; Moore, 2012; Ponton et al., 1996) . Due to the heterogeneity of behaviors and symptoms observed in children with APD (e.g., many presents with comorbid memory and attention deficits, for example), many different cortical areas, such as superior temporal, inferior parietal and inferior frontal areas involved in higher-order language and cognitive functions have been implicated (Moore, 2012 (Moore, , 2015 Poremba et al., 2003; Fritz et al., 2010) . Symptoms of APD overlap with other sensory or cognitive neurodevelopmental disorders (Sharma et al., 2009; Dawes & Bishop, 2010; AAA, 2010) . APD can cooccur with reading and language deficits in children with diagnoses of specific language impairment (SLI), reading disorder/dyslexia, autism spectrum disorder (ASD). However, this does not occur for all children (Jerger & Musiek, 2000; Sharma et al., 2009; Halliday et al., 2017; Mealings & Cameron, 2019; Dawes et al., 2009) . Additionally, children with APD can display attention (Moore et al., 2010; Gyldenkaerne et al., 2014) and/or memory problems (Sharma et al. 2009; , but many cases of APD who do not have memory or attention deficits. Thus, the overlap of cognitive, language and hearing difficulties in children with APD and other neurodevelopmental disorders is controversial for researchers who have debated whether auditory sensory processing deficits cause the listening difficulties in children diagnosed with APD (bottom-up approach: related to the ear or CANS) or cognitive deficits (top-down approach: related to cognitive function deriving from multi-modal processing) (Moore, 2012; Moore & Hunter, 2013; McFarland & Cacace, 2014; Dillon et al., 2012; Cacace & McFarland, 2013) . According to the BSA (2011) position statement, "APD is characterized by a poor perception of both speech and non-speech sounds" (p.4) and "attention is a key element of auditory processing, and that poor attention makes a major contribution to APD" (p.6). Other clinical guidelines (ASHA, 2005 ; AAA, 2010) do not specify higher-order cognitive, communicative and language-related functions in their statements as contributing factors for listening difficulties in APD (de Wit et al., 2016) . In this debate, Cacace and McFarland (2013) asserted that APD could not be considered a distinct disorder if modality specificity could not be demonstrated with certainty. Additionally, Moore and Hunter (2013) suggested APD could be distinguished from cognitive deficits based on modality specificity. Hence, the complex interactions between cognitive, language, reading, and auditory processing abilities in the brain of children with APD are not entirely understood. The potential for neuroimaging research to address this controversy and help understand of the topdown processing mechanisms in children with APD has been increasingly recognized (Moore & Hunter, 2013; Bartel-Friedrich et al., 2010; AAA, 2010; Pluta et al., 2014; Stewart et al., 2021) . In the last decade, links between atypical functional connectivity in large-scale brain organization and neurodevelopmental disorders, such as autism, Tourette's syndrome, or attention deficit hyperactivity disorder (ADHD), have been identified (Zhang & Raichle, 2010; Openneer et al., 2020; Cocchi et al., 2012; Li et al., 2014; Sadeghi et al., 2017; Faridi et al., in press ). Resting-state functional Magnetic Resonance Imaging (rsfMRI) can be utilized to explore the functionally important aspects of whole-brain intrinsic networks without requiring the participant to perform a task (van den Heuvel & Hulshoff Pol, 2010) , and hence this has been a fast-growing technique in brain imaging to study neurological brain disorders in at-risk developmental populations Fornito et al., 2013; Rosazza & Minati, 2011) . Using rsfMRI data, functional connectivity can be estimated between anatomically distributed regions by recording temporal correlations of spontaneous fluctuations in the Blood-Oxygenation-Level-Dependent (BOLD) signal (Biswal et al., 1997) . In the APD literature, to our knowledge, there have been only three resting-state fMRI studies that investigated this intrinsic activity (Pluta et al. 2014; Stewart et al., 2021, Preprint; Hoyda et al., 2021, Preprint) . The earliest of these studies investigated 13 children with diagnosed APD (without other neurological signs) and 15 healthy control (HC) children (Pluta et al., 2014) using regional homogeneity (ReHo) to investigate changes in the pattern of the default mode network (DMN). Intrinsic connectivity within this network has been associated with attentional impairment (Bonnelle et al., 2011) . The ReHo results indicated decreased functional activity in the superior frontal gyrus and posterior cingulate cortex/precuneus (Pluta et al., 2014) ; these areas are involved in control and attention and implicate the role of DMN in these cognitive processes (Castellanos et al., 2008) . Pluta et al. (2014) argued that modality-specific perceptual dysfunction could not differentiate between children with neurodevelopmental APD and those with attention deficits. More recent research involving 81 children (n=42 Listening difficulties/APD; n=39 typically developing/TD) used a region of interest (ROIs) approach to assess how speech perception and listening networks (Phonology, Intelligibility, Semantics) are different in children with APD compared to TD . Stewart and colleagues (2021) . These networks were based on differential activation of areas during contrasting listening and language processing tasks. Children with APD had increased functional connectivity in the left inferior frontal gyrus (Broca's area) and left posterior middle temporal gyrus; these regions are implicated in language production and comprehension and lexical and semantic processes (Hagoort, 2014; Acheson et al., 2013; Hickok & Poeppel, 2004) . Stewart and colleagues (2021) also reported that in the Semantics network, children with APD had stronger functional connectivity in the right parahippocampal gyrus compared to auditory areas such as left Heschl's gyrus, left middle temporal gyrus (MTG), right superior temporal gyrus (STG), and right planum temporale. Still, they had weaker connections in the left temporal fusiform cortex and right superior temporal gyrus . Stewart et al. (2021) concluded that atypical neurological aspects of APD are connected with language comprehension, consistent with the high rate of language disorder comorbidity in children with APD (Sharma et al., 2009; Dawes & Bishop, 2010; Barker et al., 2017) . A follow-up study by the same research team investigating speech and higher-order functioning in the same group of children with APD (Hoyda et al., 2021) defined ROIs for speech perception, speech production, language comprehension, naming and executive networks based on the Neurosynths database (Yarkoni et al., 2011) , and then looked for group differences in pairwise functional connectivity. Their results showed significantly decreased connectivity in the APD group compared to TD within the executive function network in the left caudate and left mid frontal gyrus suggesting a link between executive function and APD (Hoyda et al., 2021) . These studies (Pluta et al., 2014; Stewart et al., 2021; Hoyda et al., 2021) have investigated functional connectivity in children with APD based on predefined ROIs, but it is not clear yet how these ROIs and their interaction could contribute to altered brain network connectivity in APD. Thus, despite growth in the application of rsfMRI in neurodevelopmental research, the functional topological organization of children with APD remains largely unknown. Recently, graph theoretical approaches have been more widely used to analyze rsfMRI data to measure the macroscopic structural and functional attributes of brain networks (Bullmore & Sporns, 2009; Power et al., 2011) . This approach, also known as network neuroscience, offers a mathematical framework for investigating the local and global properties of neural systems (Bassett & Sporns, 2017; Sporns, 2011) , and is considered a promising tool for understanding brain networks and their behaviors (Fornito et al., 2016) . The human brain can be conceptualized as a complex network topology (Watts & Strogatz, 1998; Sporns, 2011) , which is optimally balanced in segregation (local connectivity) and integration (global connectivity) of information flow (Sporns, 2013) . These complex topologies can be represented as multiple nodes (brain regions) representing local functioning and a smaller set of nodes illustrating global functioning (Bullmore & Sporns, 2009 ). Brain organization displays scale-free, small-world (van den Heuvel et al, 2008; Bassett & Bullmore, 2006; Fornito et al., 2016) , hierarchical modularity (Meunier et al., 2010) , hub , and rich-club (van den Heuvel & Sporns, 2011) architectures. Hub architecture consists of important brain regions that interact with many other regions, facilitate functional integration, and play a key role in optimal information flow (Rubinov & Sporns, 2010) . This crucial importance of hubs makes them vulnerable spots in brain networks; an alteration in brain hubs is one of the most consistent findings in networkbased studies (Meunier et al., 2009; Rubinov & Bullmore, 2013; Crossley et al., 2014; Roger et al., 2020) . Thus investigating brain hub organization can provide new insights into the neural mechanisms underlying APD and may reveal differences in brain topologies. To the best of our knowledge, this is the first rsfMRI study that has applied graph theoretical approaches to understand the large-scale brain organization of children with APD. The present study used rsfMRI and the complex network analysis method to examine the whole brain functional topology in children diagnosed with APD and children without complaints of listening difficulties (healthy controls, HC). This research generated functional brain connectomes to measure global and brain hub topology and examine brain network differences between APD and HC. Studies of children with developmental disorders generally do not show significant differences in all global properties of the brain (Armstrong et al., 2016; Sadeghi et al., 2017; Chen et al., 2019; Zhang et al., 2019) , and hence more insights may be gained by examining regional brain topology. Based on these previous studies of children showing similarities in their global brain network organization, we hypothesized that the APD and HC groups would not differ in their resting-state functional connectivity or global topological architecture (i.e., whole-brain averaged network topology). However, children with APD showed atypical regional brain topology, specifically in cortical temporal regions associated with auditory function and hub organization. Approval for this study was granted by The University of Auckland Human Participants Ethics committee (Date: 18/10/2019, Ref. 023546). Before the start of the study, children and their parents signed assent and consent forms as per the requirement of the Ethics committee. A total of 66 children aged between 8 -14 years participated in this study, but only 57 participants remained for the analysis; nine children were excluded from the further data analysis due to excessive head motions (n = 4), incidental findings (n = 1), and uncompleted scans (n = 4). Among the 57 included participants, 28 children were already diagnosed with APD (13 boys, Mage= 10.92, range 8.58-13.41), and 29 children were healthy controls (14 boys, Mage=11.91, range 9.75-14.08). All children with APD were recruited from SoundSkills clinic in Auckland, New Zealand (https://soundskills.co.nz ). They were diagnosed formally with APD according to New Zealand guidelines for standardized testing APD test battery (Keith et al., 2019) . Children in the HC control group were recruited via posted flyers and online advertisements. They were excluded if diagnosed with hearing loss, learning difficulties, and any neuropsychiatric conditions or taking medication known to affect the central nervous system. Four children in the HC group (14%) were not experiencing any learning difficulties but had diagnoses of ASD (n=1), ADHD (n=2), or dyslexia (n=1). Twelve children with APD were also diagnosed with other comorbid developmental disorders such as dyslexia (n=8), ADHD (n=1), attention deficit disorder (ADD, n=2) and developmental language disorder (DLD, n=1). The comorbidities were allowed as these disorders coexist with APD (Dawes & Bishop 2010; O'Connor, 2012; Sharma et al., 2009 ). Children and their parents were invited to take part in two separate sessions on a single day to complete the hearing tests and the MRI scan. Most participants completed the entire procedure on the same day. All testing was conducted at the University of Auckland's Clinics and the Centre for Advanced MRI (CAMRI), located at the Grafton Campus, Auckland, New Zealand. In this session, all participants underwent 30 minutes of hearing tests, including otoscopy, pure tone air conduction audiometry (PTA), and tympanometry to check hearing acuity and screen for middle ear disease and atypical ipsilateral middle ear muscle reflexes (Roup et al., 1998) . Otoscopy results for all children were normal. PTA thresholds for every participant were no more than 20 dB HL at the octave frequencies from 0.25 to 8 kHz in each ear, and tympanogram results in both ears were indicative of normal middle ear function (static admittance in range 0.2 to 1.6 mmho, peak pressure between -100 and +20 daPa). All children were also tested on the listening-in-spatialized-noisesentences (LiSN-S; (Cameron & Dillon, 2007; 2008) test, which is a tool for assessing the ability of individuals to understand sentences in the presence of competing speech presented from different directions and using the same or different talkers. LiSN-S scores are presented as z-scores compared to normative findings for a child of the same age (Cameron et al., 2011) . The low-cue scores (speech reception threshold, SRT) reflect listening skills when no spatial or vocal cues are available to the listener to distinguish target sentences from the distractor stories. High-cue SRT scores assess listening when both vocal and spatial cues are available. Talker, Spatial, and Total Advantage scores are difference scores that reflect listeners' ability to use differences in either the voices of the speakers, the physical location of the target and distractors, or both of these types of cues, respectively, to identify the sentences in the presence of competing speech. The advantage scores are difference scores that control for variations in overall speech perception abilities across participants and accurately reflect auditory processing within the central auditory system. Specific details of the LiSN-S test are explained in Besser et al. (2015) . Participant preparation. To prepare the children for the MRI scan, a few different strategies were employed to reduce anxiety and movement during the scan session, as described in Wilke et al. (2018) . Firstly, before the appointment, all participants' parents or caregivers were encouraged to video clips to familiarize with the MRI scan procedure. Secondly, children were instructed about all the procedures using a miniature Lego scanner during their hearing session (Fig. 1) ; this helped reduce children's concerns about the scanning session. Finally, all children and their parents or caregivers were invited to attend a 20 min preparation session in the mock scanner before the actual scan started. In this session, the instructions were explained again to the children to know what would happen in the actual scan session to have a successful session. Children were instructed and practised how to stay still during the scan by playing a statue game in the mock scanner and showing them pictures illustrating how head motion can cause blurry brain images. To reduce the anxiety of children, parents or caregivers also were allowed to have the option to stay in the scan room with their children where they could see them -only a few parents chose this option. Children were allowed to bring their non-magnetic toys to the scan room. All participants were asked to use the restroom immediately before the scan as recommended by Wilke et al., (2018) . A picture of a miniature scanner model provided for this study from the Amazings team (https://www.amazings.eu/p/mri.html). This study utilized a 20-channel head coil for recording brain activity in all sequences, which is larger than the widely used 32-channel head coil. The 20-channel head coil was more comfortable for the children and allowed better control of movement using padding to fixate the . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10.1101/2022.04.05.22273478 doi: medRxiv preprint head (Greene et al., 2016) whilst accommodating the headphones needed for accessible communication. During the scan, children were asked to lay still, not fall asleep, and keep their eyes open. Children were given earplugs and headphones to minimize the loudness of the MRI noise. For the initial T1-image sequence (5 min), children were required to watch an animation of their choice to keep them entertained, while for the rsfMRI sequence (7 min; 20 s), they were asked to stay still and stare at the cross presented on the screen. Children's heads were stabilized with cushions to diminish head movements. At the end of the rsfMRI sequence, participants were asked to rate how awake they felt during the scan from 1 to 10, where 1 represents falling asleep and 10 is completely awake. MRI data were acquired on a 3T Siemens MAGNETOM Skyra scanner (Siemens, Erlangen, Germany). The high-resolution structural T1-image was acquired for co-registration using a magnetizationprepared rapid acquisition gradient echo (MPRAGE) sequence with 1 mm isotropic resolution (Field of view (FOV) = 256 mm, 208 sagittal slices in a single slab, TR = 2000 ms, Echo time (TE) = 2.85 ms, Flip angle = 8 degrees, slice thickness =1mm, acquisition time 286 sec). rsfMRI data were acquired using multi-echo/multiband (ME/MB) echo-planar imaging (EPI) sequences, customized for the pediatric population (TR 1700 ms, TEs 15, 31.63, 48.26 ms, flip angle 83 deg, multi-band factor 2, GRAPPA PAT mode, 3.2 mm isotropic voxels, 3 mm slice thickness, Field of View 202 mm, and 46 slices with full brain coverage) these sequences were adapted from Marusak et al.'s (2018) study. The Siemens embedded system automatically discards the first 10 time points of the BOLD data to achieve equilibrium, resulting in 250 volumes for the analysis. As stated in the literature, MB EPI improves temporal resolution (Preibisch et al., 2015; Demetriou et al., 2018) , and using the ME approach reduces signal dropout in brain regions and improves the temporal signal to noise ratio (tSNR) (Kundu et al., 2012; 2015) . All anatomical and functional images were converted to Nifti file sets using dcm2niix software (version 11/11/2020) developed by Rorden Lab (Rorden et al., 2007) (https://github.com/rordenlab/dcm2niix), and all Nifti files were structured according to Brain Imaging Data structure (BIDS v1.8.2, Gorgolewski et al., 2017) ( Fig. 2A) . Quality control. The functional and anatomical data quality was assessed using MRIQC's visual reports (Esteban et al., 2017) . All data were visually checked for correct alignment to the corresponding T1w image and signal artifacts. Data were inspected for head motions based on quality control measures, including carpet plots (Power, 2017) , framewise displacement (FD), DIVARS (D referring to the temporal derivative of time courses, VARS referring to RMS variance over voxels) . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10.1101/2022.04.05.22273478 doi: medRxiv preprint (Power et al., 2012; . Participants were excluded under a stringent regime based on these criteria: mean FD greater than 0.25 mm; or FDs greater than 5 mm (Parkes et al., 2018) . fMRIPrep workflow. Preprocessing of resting-state BOLD data and high-resolution T1-weighted (T1w) structural data was performed according to the fMRIPrep v20.2.3 pipeline (Esteban et al., 2019) , which is based on Nipype 1.6.1 (Gorgolewski et al., 2011) . The following description is based on the boilerplate published by the fMRIPrep pipeline, and it is covered by a "no rights reserved" (CC0) license. Internal operations of the fMRIPrep pipeline utilize a combination of software, including FSL v5.0.9 (Smith et al., 2004) , ANTs v2.3.3 (Avants et al., 2009 ), Freesurfer v6.0.1 (Dale et al., 1999 , AFNI v16.2.07 (Cox, 1996) , and Nilearn v0.6.2 (Abraham et al., 2014) . Each T1w image was corrected for intensity non-uniformity (INU) with N4BiasFieldCorrection (Tustison et al., 2010) , distributed with ANTs, and used as T1w-reference throughout the workflow. The T1w-reference was then skull-stripped with a Nipype implementation of the antsBrainExtraction.sh workflow, using the target template. Brain tissue segmentation of cerebrospinal fluid (CSF), white matter (WM), and grey matter (GM) was performed on the brainextracted T1w using FSL's fast (Zhang et al., 2001) . Volume-based spatial normalization to MNIPediatricAsym:cohort-4: res-2 (MNI's unbiased standard MRI template for pediatric data from the 7.5 to 13.5 age range) was performed through nonlinear registration with antsRegistration (ANTs), using brain-extracted versions of both T1w reference and the T1w template. For each child's functional data, first, all three echoes were slice-time corrected using AFNI's 3dTshift (Cox & Hyde, 1997) . Head-motion parameters for the BOLD reference are estimated using FSL's mcflirt (Jenkinson et al., 2002) . A B0-nonuniformity map (or field map) was also estimated based on a phase-difference map calculated with a dual-echo GRE (gradient-recall echo) sequence. The field map was then co-registered to the target EPI reference run and converted to a displacements field map with FSL's Fugue and co-registered to the T1w reference with nine degrees of freedom using FSL's FLIRT(FMRIB's Linear Image Registration Tool) (Jenkinson & Smith, 2001) with the boundary-based registration (Greve & Fischl, 2009 ) cost-function. The BOLD time-series were resampled onto their original, native space by applying a single, composite transform to correct for head-motion and susceptibility distortions. Then, the T2* map was estimated according to Posse et al.'s (1999) method and used to optimally combine BOLD data across echoes using tedana's t2smap (Kundu et al., 2012; 2015) . The BOLD time-series were then resampled into standard space (MNIPediatricAsym:cohort-4:res-2). Automatic removal of motion artifacts using independent component analysis (ICA-AROMA, Pruim et al., 2015) was performed on the functional data's time series. As a result, the "aggressive" noise-regressors were collected and placed in the corresponding confounds file (*_desc-confounds_timeseries.tsv). Several confounding time-series were calculated based on the preprocessed BOLD: FD (Jenkinson et al., 2002; Power et al., 2014) , DVARS, and three region-wise global signals. FD and DVARS are calculated for each functional run, both using their implementations in Nipype (Gorgolewski et al., 2011) . The three global signals are extracted within the CSF, the WM, . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10.1101/2022.04.05.22273478 doi: medRxiv preprint parcellation includes 300 brain regions defined based on 17 functionally parcellated networks from Yeo et al. (2011) study. Subsequently, the time series of all voxels within each region were averaged (Fig. 2C ). Next, Pearson's correlation r was computed based on the BOLD signal between all brain regions to determine pairwise functional connectivity strength. This resulted in a subject-specific symmetric and undirected weighted connectivity matrix of size 333×333. Each pair-wise element represents Pearson's correlation between the mean time courses between brain regions (Fig. 2D ). Networks are typically thresholded (removal of non-significant network edges) and binarized, but this process is semi-arbitrary (van den Heuvel et al., 2017) . Studies have shown that the density and organization of brain networks vary between subjects, which may lead to systematic changes in results depending on the threshold chosen (Wen et al., 2011) . Based on these methodological limitations, we use a matching approach of thresholding by using a range of network sparsity thresholds to ensure all subject-specific networks contain the same number of nodes and edges (Zhang et al., 2011; Achard & Bullmore, 2007; Langer et al., 2013) . We computed connectivity matrices with a network density ranging from 1 to 40% (with a 1% increment) to investigate the topological properties of the brain network. For the analysis, a minimum bound of 10% was selected to prevent graph fragmentation at a sparser threshold and an upper bound of 40% was chosen due to its liberal estimate of cerebral connectivity and its neurobiological plausibility for brain functional organization Zhang et al., 2011) (Fig. 2E ). It also ensured that all brain networks indicated network properties of small-worldness for all densities (Fig. S2 C) . For investigating the topological aspects of the functional network, graph metrics of each individual's network were calculated on the undirected weighted functional network. The graph theory computation was carried out by functions from the Brain Connectivity Toolbox (BCT v 03/03/2019, https://sites.google.com/site/bctnet/) in MATLAB R2019b (https://mathworks.com/). Functional integration is defined as combining specialised information across nodes and shows how the network can share the information in the distributed nodes (Rubinov & Sporns, 2010) . Measures of characteristic path length (CPL) and global efficiency (Eglob) were calculated to estimate network integration. A short path length indicates that each node can reach other nodes with few steps (i.e., the minimum number of edges connecting two connected brain regions), or a path composed of few edges; the average of the shortest path between all pairs of nodes is defined as CPL (Rubinov & Sporns, 2010) . Eglob is also a measure that indicates the capacity of a network for transferring parallel information in each pair of nodes, and it is defined as the inverse of the shortest path length between nodes (Latora & Marchiori, 2001) . Therefore CPL and Eglob are primary metrics to measure network integration (Achard & Bullmore, 2007) . Functional segregation refers to the ability in the brain that specialized processing happens in the densely interconnected group of brain regions (Rubinov & Sporns, 2010) . To estimate the local properties of brain networks, a range of metrics such as mean local efficiency (Eloc), Clustering Coefficient (CC), and Modularity Optimization (Mod) were calculated. Eloc metric . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10.1101/2022.04.05.22273478 doi: medRxiv preprint measures the network capability in transferring the information at the local level (node's neighborhood). It is defined as the average inverse shortest path between two nodes (Rubinov & Sporns 2010; Latora & Marchiori, 2001) . CC is defined as the fraction of triangular connected nodes around the node of interest and indicates the degree to which neighbors tend to cluster with each other in the network (i.e., Cliquiness; Watts & Strogatz, 1998) , hence the mean CC is considered as a direct measure of brain segregation (Rubinov & Sporns 2010) . Mod is a segregation measure that indicates the presence of densely interconnected nodes (i.e., modules) and indicates the size and composition of these modular structures (Girvan & Newman, 2002; Rubinov & Sporns, 2010) . Smallworldness (SW) is an analogy based on the small-world phenomenon (Watts & Strogatz, 1998) , indicating that a biological system such as the brain has a structure that is neither regular or random (Latora & Marchiori, 2001) . This property is highly clustered like a regular lattice network but has a similar characteristic path length to a random network (Watts & Strogatz, 1998) . Hence, the SW metric can present the segregation and integration of the network and shows a high global and local efficiency of a complex system (Pievani et al., 2011) . To examine the SW property of each individual's brain network, 100 random networks (BCT, randomio_und function) with the same topological properties of empirical networks and a wiring cost of 20 were generated based on the Maslov-Sneppen Null network model that preserves the degree distribution of the original networks (Maslov & Sneppen, 2002 ) (See Fig. 2F ). Hub detection. Important nodes or hubs participate in many interactions within a network and play a key role in optimal information flow in the brain networks (Rubinov & Sporns 2010; Stam, 2014; Power et al., 2013) . As proposed by Guimerà and Nunes Amaral (2005) , hubs and their topological roles can be quantified by measures of within-module degree z-score (WMZ) and participation coefficient (PC). WMZ is a measure that estimates how strongly a node is connected to its module relative to other nodes within the same community (intra-modular connections). PC is an intermodular degree that describes if nodes are distributed uniformly across many modules or are concentrated within a module. This metric reveals the modular segregation and intermodular integration of the network in the form of the provincial hub (nodes with high WMZ, low PC) and connector hubs (nodes with high WMZ and high PC) (Rubinov & Sporns, 2010; Fornito et al., 2016; Guimerà & Nunes Amaral, 2005; Meunier et al., 2009) . To perform hub detection, first, group average matrices were calculated for each group at the top 10% of the connections because of the high rank of their correlation strength (Alavash et al., 2019) (Fig. 2H) . Afterwards, the Louvain community detection algorithm (BCT: community_louvain.m) was applied to reveal the modular brain structure (Blondel et al., 2008) due to its fast and reliable procedure in detecting communities (Betzel, 2020) . Since this algorithm is heuristic and stochastic, this process was repeated 1000 times to reach a consensus for determining community labels (concencus_und.m, Betzel, 2020) (Fig. 2I) . Once all nodes were assigned to their modules, WMZ (module_degree_zscore.m) and PC were calculated. Recent work by Pedersen and colleagues (2020) has shown that to reduce the influence of intramodular connectivity, which may result in inaccurate inference in a network with different modular sizes and also maximize the identification of interconnected nodes, it is beneficial to normalize PC with randomly generated null networks (PCnorm; Pedersen et al., 2020) . Hence in this study PCnorm was computed by participation_coef_norm.m . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10.1101/2022.04.05.22273478 doi: medRxiv preprint (https://github.com/omidvarnia/Dynamic_brain_connectivity_analysis/) function based on 1000 randomizations which were previously shown as an adequate iteration for estimating stable PCnorm (Pedersen et al., 2020) (See Fig. 2J ). Lastly, brain hubs and their role in networks were identified based on the criteria described according to an earlier study by Meunier and colleagues (2009) as follows: connector hub: WMZ>1 and PCnorm >0.5; provincial hub: WMZ>1 and PCnorm <0.5. Group characteristics and behavioral variables were tested with two-sample t-tests to analyze differences in age, head motion (mean FD), and LiSN-S results. Pearson chi-square tests were also used to compare handedness and gender distribution between groups. To assess the group differences across the entire connectome, we used the network-based statistics toolbox (NBS, https://sites.google.com/site/ bctnet/comparison/nbs) (Zalesky et al., 2010b) . The NBS is a connectome-wide analysis approach that improves statistical power over common mass-univariate correction methods such as FDR by localizing the strength of connectivity to identify significant effects at the components level (subnetworks) (Zalesky et al., 2010b; Fornito et al., 2016) . In NBS, these subnetworks are constructed by applying an initial threshold to the data (p<0.05, uncorrected). The observed subnetwork sizes were then compared with the empirical null distribution of maximal component sizes acquired by the permutation test (10000 times) (Fornito et al., 2016) . This evaluation of observed component sizes with regards to the null distribution of maximal sizes controls for the Family-wise error (FWE) for resulting inference on subnetworks (Fornito et al., 2016) . First, all individuals' connectivity matrices were Fisher's R to Z transformed (Mudholkar, 2014) . Two-sample t-tests were then performed for each pairwise connection linking 333 brain regions to test group differences in functional connectivity in either direction (two-tailed hypothesis test). Age was included as a covariate for these analyses. Differences in hub measures. To assess whether there are group differences in hub detection measures of PC, PCnorm, and WMZ, the area under the curve (AUC) across the sparsity range of 10 to 40% was calculated. The AUC was chosen because integrating network density costs could control the monotonic transformation of a weight set of the weighted graph (Ginestet et al., 2011) and it may improve the sensitivity in detecting the case-control group differences (Achard & Bullmore, 2007; Hosseini et al., 2012; van den Heuvel et al., 2017) . Two-sample t-tests assuming unequal variances between APD and HC groups were carried out on the AUC of each network measure in the Permutation Analysis of Linear model software (PALM; https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/PALM/ UserGuide). PALM uses a permutation test in a general linear model framework while controlling for false-positive errors under weak or unreasonable assumptions. The data are primarily exchangeable under the null hypothesis (Winkler, Ridgway, Webster, Smith, Nichols., 2014) . To test the null hypothesis that the group means were all equal, for each network measure, all values randomly were assigned to all subjects and the mean differences were recomputed between two randomized groups. The randomization was repeated 20000 times, and the 95% confidence interval was calculated and used as the critical value of significance testing (p<0.05). The effect of age as a . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10.1101/2022.04.05.22273478 doi: medRxiv preprint nuisance confound was also controlled during the randomization (demeaned). To control the multiple comparisons problem, the p-value of all network measures was corrected using false discovery rate (FDR) correction using an alpha level of 0.05 (Benjamini & Hochberg, 1995) (See Fig. 2G ). Exploratory partial correlation analysis was performed to assess the association between nodal measures of PC, PCnorm and WMZ with LiSN-S results (z-scored) using a permutation test (20000 randomizations) and controlling the effect of age (p values corrected with FDR correction, p<0.05). For this analysis, three participants (HC=1, APD=2) were excluded due to missing behavioral data for correlation analysis between the brain and LiSN-S data. Meta-analytic interpretation. Neurosynth (Yarkoni et al., 2011) , a data-driven tool for functional meta-analysis, was utilized for data interpretation. The Neurosynth database contains more than 14000 functional neuroimage studies that allow the identification of associated cognitive terms with the brain activation pattern. Neurosynth extracted and visualised the top 20 cognitive terms (excluding redundant, anatomical terms) associated with significant ROIs as a word cloud in MATLAB R2019b. After identifying ROIs showing significant group differences, the Neurosynth database was searched for these ROIs to identify relevant cognitive terms. For each significant region, up to 20 associated cognitive terms were selected and visualized for each ROI based on their meta-analytic coactivation scores (Pearson r, uncorrected). . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10.1101/2022.04.05.22273478 doi: medRxiv preprint Figure 2 Schematic representations of study pipeline. First, all participants' raw functional data were preprocessed and denoised (A) and then parcellated into 333 ROIs using Gordon parcellation (B), corresponding time series were extracted and averaged within each ROI to calculate individual connectivity matrices (D), afterwards, each connectivity matrix was thresholded and graph theory of global and nodal measures were calculated (F) and then area under the curve at sparsity of 10 to 40% was computed for statistical analysis with the permutation test (G). Also, the group averaged matrices were calculated for each group at the network density of 10% (H) and using the Louvain algorithm modular architecture of each group was revealed (I) to identify hubs and their functional roles based on WMZ and PCnorm metric (J). Group comparisons of demographic variables, including motion (mean FD), handedness, and gender, showed no significant differences. However, APD and HC were significantly different in age (HC>APD). Group comparisons of LiSN-S scores only indicated significant differences in Talker Advantage scores (HC>APD), although Fig. 3 shows poorer advantage scores on average for the APD group. Further demographic details are depicted in Table 1. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10.1101/2022.04.05.22273478 doi: medRxiv preprint Note: Data are presented as n or mean ± standard deviation. HC, healthy control; APD, auditory processing disorder; FD, framewise displacement; LiSN-S, Listening in Spatialized Noise -Sentences Test. Between-group differences were tested with two-sample t-tests a and Pearson chi-square tests b . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10.1101/2022.04.05.22273478 doi: medRxiv preprint Figure 3 Distribution of LiSN-S z-scores (Total advantage, Spatial advantage, Talker advantage, Highcue, Low-cue) between APD and HC groups 3.2 Edge-wise functional connectivity NBS assessed group differences in functional connectivity. The results showed no significant difference in edge-wise connectivity between APD and HC groups after controlling for multiple comparisons (FWR-corrected). Investigating network segregation indicated higher mean CC and mean Eloc in the APD group than HC. However, group comparisons revealed no statistical differences across the range of network density thresholds (1 to 40%). Fig. S2 presents results from Gordon parcellation with 333 functionally separated regions. Outcomes from other global measures are provided in supplementary materials. Examining the modular organization of HC and APD groups (Density 10%), there were similarities (Fig. 4) but also distinct differences regarding the size and topological role of community structures between groups. The modular community detection analysis revealed five functional modules in both . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10.1101/2022.04.05.22273478 doi: medRxiv preprint groups, including default mode-ventral attention (Fig. 4 Module Ⅰ, Blue), somatomotor ( Fig. 4 Module Ⅱ, Red), limbic ( Fig. 4 Module Ⅲ, Orange), visual ( Fig.4 Module Ⅳ, Purple), and frontoparietal-dorsal attention (Fig. 4 Module Ⅴ, Green) . The most extensive module in both groups was default mode-ventral attention, with 86 ROIs in APD and 90 ROIs in HC groups, comprising the default mode (DM), cinguloparietal, and ventral attention regions. The smallest module in both groups was somatomotor with 50 and 38 ROIs, respectively (See Table 2 ). Fig. 5 displays brain hubs and their role in the network (i.e., connector or provincial) based on WMZ and PCnorm. A major difference between the two groups was driven by the somatomotor module, where APD had 10 provincial hubs compared to HC with 4 provincial hubs. Also, the APD group had several regions where nodes were identified as hubs, in contrast to HC where these regions were not hubs, in the DM (ROIs #6, #162), somatomotor-hand (ROIs #2, #50, #57, #163), and dorsal attention (DA; ROI #74) regions. More details about group differences in hub regions are presented in Table S1 . Results from the group analysis based on the nodal measure of PC revealed a significant group difference in the region of DMN. This was observed in the right STG in node #331 (DMN, p=0.0333, t=3.985, FDR corrected). Fig. 6 illustrates the result for the PC measure (AUC 10-40%) of 333 functionally separated brain regions. The red color indicates an elevation of the PC metric in the HC group (APD0.5, WMZ<1) in the visual module that facilitates interactions between modules/networks. This contrasts with Kaboodvand et al.'s (2018) study of healthy adults, which suggested that RT is a provincial hub within DMN; such a difference in results could be partly related to age differences in the study populations. Our group analysis based on Schaefer parcellation also indicated differences in the bilateral pTO (ROIs #56, #298) (Fig. 6B) . Activation of bilateral Temporo-occipital cortices is associated with facial expression, visual motion, action observation, emotional expressions (Fig. 8B) and has been previously explored in reading difficulties and in facial processing difficulties in people with epilepsy (Tailby et al., 2014) and ASD (Herbert et al., 2002; Koshino et al., 2008) . Although still relatively unexplored, visual processing deficits have been reported in children with APD, for example, Dawes et al.'s (2009) study of children with APD and comorbid dyslexia found visual processing difficulties as part of a multifactorial description of learning problems in these children. Furthermore, in our group comparison, we found a significant PC difference in the left IPS (ROI #93), and right posterior INS (ROI #216). The IPS region is situated in the superior parietal lobe and is thought to be highly associated with visuomotor tasks, visuospatial attention (Grefkes & Fink 2005; Silk et al., 2010; Materna et al., 2008) , and working memory (Sharma et al., 2009; Gokula et al., 2019) (Fig. 8B ). This outcome is consistent with reports of the co-occurrence of APD with ADHD symptoms (Chermak et al., 1998; 2002) . The anterior INS is thought to be connected to a variety of human social emotions such as compassion, empathy, taste, fear, pleasure, and disgust (Lamm & Singer 2010; Mutschler et al., 2009; Ostrowsky et al. 2002; Chen et al., 2014) or homeostatic emotions such as pain (Ostrowsky et al., 2002) (Fig. 8B) . However, neuroimaging studies of posterior INS have also revealed its connection to auditory cortices (Zhang et al., 2019; Dennis et al., 2014; Cloutman et al., 2012) and suggest the involvement of posterior INS with auditory processing, including allocating auditory attention, phonological processing (Bamiou et al., 2003; Remedios et al., 2009) , and auditory responses observed from Heschl's gyrus (Zhang et al., 2019) . . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10.1101/2022.04.05.22273478 doi: medRxiv preprint Although our research provides insights into the brain organization of children with APD, several methodological issues should be addressed in future studies. Collecting a large sample size in a single small country like New Zealand during the global pandemic was challenging for this study. Although our sample size was, in theory, large enough to elicit reliable graph theory measures according to previous studies (Termenon et al., 2016; Andellini et al., 2015) , we believe future research would benefit from data sharing approaches to gather larger data cohorts to validate our results as a generalizable finding. We observed that the choice of brain parcellation affected our results with more significant regions in the Schaefer parcellation than the Gordon parcellation (as the number of nodes affects the interpretation of fMRI time-series and statistical comparisons). The goal of parcellation is to derive a set of homogenous brain regions with high intra-node connectivity -this is an ill-posed issue in network science that is known to affect results. This semi-arbitrary selection of parcellation and its effect on global and local properties of brain network has been discussed for a long time among researchers in the field (Zalesky et al., 2010a; Andellini et al., 2015; Termenon et al., 2016; Arslan et al., 2018; Bryce et al., 2021) , and the solution that has been recommended is to apply multiple parcellation schemes to understand the influence of brain parcellation (Bryce et al., 2021) . Evidence from numerous neuroimaging studies indicates that the human connectome is unique to individuals in terms of connectional traits (Mueller et al., 2013; Barch et al., 2013; Gordon et al., 2017) , where these distinct features can be distinguished among individuals (Finn et al., 2015) . Also, these unique identified patterns of brain networks could change in different tasks (Salehi et al., 2019) . Since group-wise parcellations (e.g., Gordon or Schaefer parcellations) may not show these unique individual features and patterns, we hope that future research applies individualized parcellation to improve the understanding of APD's functional organization, which would result in a deeper understanding of brain activity associated with cognitive and listening behaviors in APD. In addition to using conventional methods where brain regions (i.e., Nodes) and their functional interactions (i.e., Edges) are the focus (Node-centric), network neuroscience has proposed a new approach for investigating the functional organization of the human brain that is centered on studying edge interactions (Edge-centric) (Faskowitz et al., 2020; Betzel, 2020; Esfahlani et al., 2021) . The Edge-centric method relies on bivariate Pearson correlation (for quantifying functional connectivity) and utilizes co-fluctuations time courses between pairs of nodes instead of averaging them (Betzel, 2020) . By improving temporal resolution, Edge-centric models could reveal new aspects of brain networks' functional or structural topology (Faskowitz et al. 2020; . Another issue that arose from our cross-sectional study was that although our APD sample was diagnosed based on a comprehensive diagnostic assessment protocol (28 children diagnosed with APD), the sample was heterogeneous and included a range of children who had been recently . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10.1101/2022.04.05.22273478 doi: medRxiv preprint diagnosed (39%) or diagnosed more than a year previously (60%). Some children were under treatment (32%). Future research can minimize this by utilizing a longitudinal study design to investigate changes in brain functional organization from diagnosis until treatment. Additionally, APD commonly occurs with other comorbid disorders (Sharma et al., 2009; Dawes & Bishop 2010; O'Connor, 2012; Burns, 2013; Gokula et al., 2019) . Children diagnosed with APD indicate symptoms of listening difficulties when they are referred to audiology clinics, and these difficulties can be caused by auditory processing, language, cognitive, or hearing deficits (Dawes et al., 2008; Dawes & Bishop, 2010; Moore, 2012; Ankmnal-Veeranna et al., 2019; Magimairaj & Nagaraj, 2018; Dillon & Cameron, 2021) . These prevalent comorbid conditions were also identified in our study populations. The participant diagnostic reports showed that almost half of the children (12 out of 28) had comorbid disorders such as dyslexia (8/29%), ADHD/ADD (3/11%), and DLD (1/3%), and the others were only diagnosed with APD (16/57%). Although we were able to see group differences in brain regions, a larger sample that separates children with APD with and without different comorbidities could better unravel differences in brain processing between these subgroups. We suggest that future studies investigate separate groups of children based on their diagnosis. This differentiation could help improve the understanding of how the functional organization of APD has been affected by comorbid disorders. Despite a couple of reports on the structural connectivity of APD (Schmithorst et al., 2013; Farah et al., 2014) , the brain structural organization of children with APD has not been fully explored. We recommend future studies by acquiring diffusion-weighted MRI data and applying network neuroscience to investigate WM microstructural organization of APD individuals to help understand neural pathways involved in listening difficulties. In summary, this study presents new evidence of changes in brain network function in children with APD. Despite similar modular organization and whole-brain connectivity in both groups, children with APD showed atypical hub architecture compared to HC. Our results also indicated that the alteration of the brain topology in APD exists at the regional level, not at the global level. The group comparison based on the PC metric revealed significant differences in the right and bilateral auditory areas (e.g., STG) according to two methodologically different parcellations. These areas are associated with auditory processing, listening, speech, language, emotions and memory perception within DMN, indicating the role of multi-modal factors for listening difficulties in APD children. Our study underlines the importance of future research utilizing network science to understand the neural bases of APD and the use of these regional biomarkers as a potential clinical support tool. Auckland for assisting in participant recruitment. We would also like to acknowledge Dr. Catherine Morgan for her contribution in modifying MRI data acquisition protocols and Meghan van der Meer for her assistance in editing the final manuscript. Further we would like to thank the MRI technicians at CAMRI for their continued support in assisting with research studies and conducting these scans. This study was supported by Eisdell Moore Center for Hearing and Balance Research (Grant No. 3716796 ). The authors have no conflict of interest to declare. The data that support the findings of this study are available from the corresponding author upon reasonable request. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10. 1101 /2022 Resonance In Vivo, , 171-178. https://doi.org/10.1002/(SICI)1099-1492(199706/08) 10:4/5<171::AID-NBM453>3.0.CO;2-L Crossley, N. A., Mechelli, A., Scott, J., Carletti, F., Fox, P. T., McGuire, P., & Bullmore, E. T. (2014) . The hubs of the human connectome are generally implicated in the anatomy of brain disorders. Brain, 137 (8) Comparing resting state fMRI de-noising approaches using multi-and single-echo acquisitions. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10.1101/2022.04.05.22273478 doi: medRxiv preprint Zhang, Y., Zhou, W., Wang, S., Zhou, Q., Wang, H., Zhang, B., ... & Wang, X. (2019) . The roles of subdivisions of human insula in emotion perception and auditory processing. Cerebral Cortex, 29(2), 517-528. https://doi.org/10.1093/cercor/bhx334 Additional supporting information may be found online in the supporting Information section at the end of this article. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 6, 2022. ; https://doi.org/10.1101/2022.04.05.22273478 doi: medRxiv preprint Machine learning for neuroimaging with scikit-learn Efficiency and cost of economical brain functional networks Stimulating the brain's language network: syntactic ambiguity resolution after TMS to the inferior frontal gyrus and middle temporal gyrus Neural bases of accented speech perception Modular reconfiguration of an auditory control brain network supports adaptive listening behavior The resting human brain and motor learning Auditory processing disorders: Relationship to cognitive processes and underlying auditory neural integrity Diagnosis, treatment, and management of children and adults with central auditory processing disorder Central auditory processing: Current status of research and implications for clinical practice Central) auditory processing disorders Test-retest reliability of graph metrics of resting state MRI functional brain networks: A review Functionalanatomic fractionation of the brain's default network Auditory brainstem responses in children with auditory processing disorder Graph-theoretical analysis of resting-state fMRI in pediatric obsessive-compulsive disorder Human brain mapping: A systematic comparison of parcellation methods for the human cerebral cortex Advanced normalization tools (ANTS) The insula (Island of Reil) and its role in auditory processing: literature review Function in the human connectome: task-fMRI and individual differences in behavior Cortical auditory-evoked potential and behavioral evidence for differences in auditory processing between good and poor readers Development of fMRI tests for children with central auditory processing disorders Small-world brain networks Network neuroscience FMRI study of emotional speech comprehension Investigations into resting-state connectivity using independent component analysis Controlling the false discovery rate: a practical and powerful approach to multiple testing The diverse club A mechanistic model of connector hubs, modularity and cognition Speech-in-speech listening on the LiSN-S test by older adults with good audiograms depends on cognition and hearing acuity at high frequencies Community detection in network neuroscience Network neuroscience and the connectomics revolution Human temporal lobe activation by speech and nonspeech sounds A comparison of five fMRI protocols for mapping speech comprehension systems Simultaneous assessment of flow and BOLD signals in resting-state functional connectivity maps. NMR in Biomedicine: An International Journal Devoted to the Development and Application of Magnetic Resonance In Vivo Fast unfolding of communities in large networks Default mode network connectivity predicts sustained attention deficits after traumatic brain injury Position Statement: Auditory processing disorder (APD). 1-9 Brain parcellation selection: An overlooked decision point with meaningful effects on individual differences in resting-state functional connectivity The brain's default network: anatomy, function, and relevance to disease Complex brain networks: graph theoretical analysis of structural and functional systems Auditory Processing Disorders and Literacy Factors influencing tests of auditory processing: A perspective on current issues and relevant concerns Development of the listening in spatialized noise-sentences test (LISN-S) The Listening in Spatialized Noise-Sentences Test (LISN-S): Comparison to the prototype LISN and results from children with either a suspected (central) auditory processing disorder or a confirmed language disorder Listening in Spatialized Noise-Sentences Test (LiSN-S): Normative and retest reliability data for adolescents and adults up to 60 years of age Cingulate-precuneus interactions: a new locus of dysfunction in adult attentiondeficit/hyperactivity disorder Anterior insular cortex activity to emotional salience of voices in a passive oddball paradigm. Frontiers in human neuroscience Disrupted brain functional networks in drug-naïve children with attention deficit hyperactivity disorder assessed using graph theory analysis Tinnitus distress is associated with enhanced resting-state functional connectivity within the default mode network Behavioral signs of central auditory processing disorder and attention deficit hyperactivity disorder Differential diagnosis and management of central auditory processing disorder and attention deficit hyperactivity disorder Behavioral characteristics of auditory processing disorder and attention-deficit hyperactivity disorder: predominantly inattentive type Mitigating head motion artifact in functional connectivity MRI The variation of function across the human insula mirrors its patterns of structural connectivity: evidence from in vivo probabilistic tractography Altered functional brain connectivity in a non-clinical sample of young adults with attentiondeficit/hyperactivity disorder The segregation and integration of distinct brain networks and their relationship to cognition Multi-task connectivity reveals flexible hubs for adaptive task control AFNI: software for analysis and visualization of functional magnetic resonance neuroimages Software tools for analysis and visualization of fMRI data Imaging-based parcellations of the human brain Edge-centric analysis of time-varying functional brain networks with applications in autism spectrum disorder MRIQC: Advancing the automatic prediction of image quality in MRI from unseen sites fMRIPrep: a robust preprocessing pipeline for functional MRI Altered white matter microstructure underlies listening difficulties in children suspected of auditory processing disorders: a DTI study Brain Structural Correlates of INTELLIGENCE in ADHD Individuals. Basic and Clinical Neuroscience Edge-centric functional network representations of human cerebral cortex reveal overlapping system-level architecture Edges in brain networks: Contributions to models of structure and function Subtyping cognitive profiles in autism spectrum disorder using a functional random forest algorithm Cerebral representation of one's own past: neural networks involved in autobiographical memory Functional connectome fingerprinting: identifying individuals using patterns of brain connectivity Network scaling effects in graph analytic studies of human resting-state FMRI data. Frontiers in systems neuroscience, 4, 22 What can spontaneous fluctuations of the blood oxygenationlevel-dependent signal tell us about psychiatric disorders Functional dysconnectivity of corticostriatal circuitry as a risk phenotype for psychosis Connectomics: a new paradigm for understanding brain disease The connectomics of brain disorders Fundamentals of brain network analysis Functional connectivity: the principal-component analysis of large (PET) data sets Adaptive, behaviorally gated, persistent encoding of task-relevant auditory information in ferret frontal cortex Brain network analysis: separating cost from topology using cost-integration Comorbidity of auditory processing, attention, and memory in children with word reading difficulties Generation and evaluation of a cortical area parcellation from resting-state correlations Individual-specific features of brain systems identified with resting state functional correlations Three distinct sets of connector hubs integrate human brain function Default-mode network streams for coupling to language and control systems Nipype: a flexible, lightweight and extensible neuroimaging data processing framework in python BIDS apps: Improving ease of use, accessibility, and reproducibility of neuroimaging data analysis methods Control networks and hubs Considerations for MRI study design and implementation in pediatric and clinical populations The functional organization of the intraparietal sulcus in humans and monkeys Accurate and robust brain image alignment using boundary-based registration Functional cartography of complex metabolic networks Attend to this: The relationship between auditory processing disorders and attention deficits Image processing and analysis methods for the Adolescent Brain Cognitive Development Study Nodes and networks in the neural architecture for language: Broca's region and beyond Auditory processing deficits are sometimes necessary and sometimes sufficient for language difficulties in children: Evidence from mild to moderate sensorineural hearing loss Graph theory approaches to functional network organization in brain disorders: A critique for a brave new small-world Abnormal asymmetry in language association cortex in autism Dorsal and ventral streams: a framework for understanding aspects of the functional anatomy of language Prevalence of clinical referrals having hearing thresholds within normal limits Assistive listening devices drive neuroplasticity in children with dyslexia GAT: a graph-theoretical analysis toolbox for analyzing between-group differences in large-scale structural and functional brain networks Resting State fMRI Speech, Language, and Executive Function Network Connectivity in Children with and without Listening Difficulties Current evidence does not support the claims made for CogMed working memory training Improved optimization for the robust and accurate linear registration and motion correction of brain images A global optimisation method for robust affine registration of brain images Report of the consensus conference on the diagnosis of auditory processing The retrosplenial cortex: A memory gateway between the cortical default mode network and the medial temporal lobe The Buffalo CAPD Model: The importance of phonemes in evaluation and remediation New Zealand guidelines on auditory processing disorder A dual-subsystem model of the brain's default network: self-referential processing, memory retrieval processes, and autobiographical memory retrieval A role for the right superior temporal sulcus in categorical perception of musical chords fMRI investigation of working memory for faces in autism: visual coding and underconnectivity with frontal areas Distinct functional substrates along the right superior temporal sulcus for the processing of voices Differentiating BOLD and non-BOLD signals in fMRI time series using multi-echo EPI Robust resting state fMRI processing for studies on typical brain development based on multiecho EPI acquisition The role of anterior insular cortex in social emotions The problem of thresholding in small-world network analysis Efficient behavior of small-world networks Network-based analysis reveals stronger local diffusion-based connectivity and different correlations with oral language skills in brains of children with high functioning autism spectrum disorders European 17 countries consensus endorses more approaches to APD than reported in Wilson Working memory and auditory processing in school-age children The frontoparietal network: function, electrophysiology, and importance of individual precision mapping Poor between-session recall of extinction learning and hippocampal activation and connectivity in children Specificity and stability in topology of protein networks Dissociable roles of the superior temporal sulcus and the intraparietal sulcus in joint attention: a functional magnetic resonance imaging study Modality specificity is the preferred method for diagnosing the auditory processing disorder (APD): Response to Moore and Ferguson Investigating auditory spectral and temporal resolution deficits in children with reading difficulties Age-related changes in modular organization of human brain functional networks Modular and hierarchically modular organization of brain networks Processing of natural sounds in human auditory cortex: tonotopy, spectral tuning, and relation to voice sensitivity Nature of auditory processing disorder in children The diagnosis and management of auditory processing disorder. Language, Speech and Hearing in schools Listening difficulties in children: Bottom-up and top-down contributions Auditory processing disorder (APD) in children: a marker of neurodevelopmental syndrome. Hearing, Balance and Communication Evolving concepts of developmental auditory processing disorder (APD): a British Society of Audiology APD special interest group 'white paper Sources of pathology underlying listening disorders in children Listening difficulties in children: Behavior and brain activation produced by dichotic listening of CV syllables Fisher's z-Transformation Individual variability in functional connectivity architecture of the human brain Functional organization of the human anterior insular cortex Auditory processing in autism spectrum disorder: a review The development of brain network hubs Early and late development of hub connectivity in the human brain. Current opinion in psychology A graph theory study of resting-state functional connectivity in children with Tourette syndrome Representation of pain and somatic sensation in the human insula: a study of responses to direct electrical cortical stimulation Anatomy of the auditory cortex An evaluation of the efficacy, reliability, and sensitivity of motion correction strategies for resting-state functional MRI Reducing the influence of intramodular connectivity in participation coefficient Damage to the default mode network disrupts autobiographical memory retrieval Differential remoteness and emotional tone modulate the neural correlates of autobiographical memory Functional network disruption in the degenerative dementias Reduced resting-state brain activity in the default mode network in children with (central) auditory processing disorders Auditory brainstem response generation by parallel pathways: differential maturation of axonal conduction time and synaptic transmission Functional mapping of the primate auditory system Enhancement of BOLD-contrast sensitivity by single-shot multi-echo functional MR imaging Functional network organization of the human brain Spurious but systematic correlations in functional connectivity MRI networks arise from subject motion Evidence for hubs in human functional brain networks Methods to detect, characterize, and remove motion artifact in resting state fMRI A simple but useful way to assess fMRI scan qualities Evaluation of multiband EPI acquisitions for resting state fMRI ICA-AROMA: A robust ICA-based strategy for removing motion artifacts from fMRI data Measuring perceptions of classroom listening in typically developing children and children with auditory difficulties using the LIFE-UK questionnaire A default mode of brain function An auditory region in the primate insular cortex responding preferentially to vocal communication sounds Improving lesion-symptom mapping Resting-state brain networks: literature review and clinical applications Tympanometric screening norms for adults Complex network measures of brain connectivity: uses and interpretations Schizophrenia and abnormal brain network hubs Screening of autism based on task-free fmri using graph theoretical approach There is no single functional atlas even for a single individual: Functional parcel definitions change with task Impact of in-scanner head motion on multiple measures of functional connectivity: relevance for studies of neurodevelopment in youth Local-global parcellation of the human cerebral cortex from intrinsic functional connectivity MRI Default mode, dorsal attention and auditory resting state networks exhibit differential functional connectivity in tinnitus and hearing loss Left ear advantage in speech-related dichotic listening is not specific to auditory processing disorder in children: A machine-learning fMRI and DTI study Comorbidity of auditory processing, language, and reading disorders Attention, memory, and auditory processing in 10-to 15-year-old children with listening difficulties The dynamics of functional brain networks: integrated network states during cognitive task performance Spatial working memory and spatial attention rely on common neural processes in the intraparietal sulcus Advances in functional and structural MR image analysis and implementation as FSL The human connectome: a complex network Network attributes for segregation and integration in the human brain Modern network science of neurological disorders Speech cortical activation and connectivity in typically developing children and those with listening difficulties Reading difficulty is associated with failure to lateralize temporooccipital function Reliability of graph analysis of resting state fMRI using test-retest dataset from the Human Connectome Project Lateralization of auditory-cortex functions Retrosplenial cortex is required for the retrieval of remote memory for auditory cues Mapping aesthetic musical emotions in the brain N4ITK: improved N3 bias correction Small-world and scale-free organization of voxel-based resting-state functional connectivity in the human brain Rich-club organization of the human connectome The influence of head motion on intrinsic functional connectivity MRI GRETNA: a graph theoretical network analysis toolbox for imaging connectomics Parcellating cortical functional networks in individuals Collective dynamics of 'small-world' networks Discrete neuroanatomical networks are associated with specific cognitive abilities in old age Clinical application of advanced MR methods in children: points to consider Evolving the concept of APD BrainNet Viewer: a network visualization tool for human brain connectomics Large-scale automated synthesis of human functional neuroimaging data The organization of the human cerebral cortex estimated by intrinsic functional connectivity Whole-brain anatomical networks: does the choice of nodes matter? Network-based statistic: identifying differences in brain networks Structure and function of auditory cortex: music and speech Disease and the brain's dark energy Disrupted brain connectivity networks in drug-naive, first-episode major depressive disorder Altered brain functional network in children with type 1 Gaucher disease: a longitudinal graph theory-based study Segmentation of brain MR images through a hidden Markov random field model and the expectation-maximization algorithm We would like to thank all the participants and their parents for participating in this study during the COVID-19 global pandemic. We also thank Dr. Bill Keith, Mrs. Sam Marsh and Soundskill clinic in