The effects of exposure to road traffic noise at school on central auditory pathway functional connectivity

As the world becomes more urbanized


Introduction
Within our urbanized world, exposure to noise has become unavoidable, especially in highly urbanized areas and major cities where demand of transport and road traffic has suffered a great growth during the last years (European Environmental Agency, 2020).Environmental noise has been associated with a range of different auditory and non-auditory adverse health effects (Basner et al., 2014;Basner and McGuire, 2018;Eggermont, 2017;Kempen et al., 2018;World Health Organization, 2018) including annoyance (Guski et al., 2017;Ohrstrom et al., 2006), sleep disturbance (Basner and McGuire, 2018;Smith et al., 2022), cardiovascular diseases (Kempen et al., 2018) and mortality Abbreviations: CSF, cerebrospinal fluid; dB, decibels; EPI, echoplanar imaging; ISO, International Standards Organization; LAeq, A-weighted equivalent sound pressure levels; MNI, Montreal Neurological Institute; MRI, Magnetic Resonance Imaging; fMRI, functional Magnetic Resonance Imaging; FEW, family-wise error; ROI, regions of interest; SPM, Statistical Parametric Mapping; WHO, World Health Organization.(Cole-Hunter et al., 2022), among others (Clark et al., 2020).Due to its public health relevance, the World Health Organization (WHO) has been developing recommendations and guidelines with the aim to limit the levels of environmental noise exposure and protect human health (World Health Organization, 2018).
Focusing on the possible harmful effects produced in children, road traffic noise in school environment has been associated with poorer mental health, although the evidence is still limited (Clark and Paunovic, 2018;Clark et al., 2020), and with poorer cognitive performance and development (Clark et al., 2020;Clark and Paunovic, 2018;Erickson and Newman, 2017;Stansfeld et al., 2005;Stansfeld and Clark, 2015;Matheson et al., 2010;Foraster et al., 2022).These studies have mostly focused on noise exposure at school, where children spend most of the weekdays and concentration occurs.
To our knowledge, there is only one functional imaging study exploring the possible impact of road traffic noise on brain in schoolchildren (Pérez-Crespo et al., 2022).In this functional connectivity MRI approach, Pérez-Crespo et al. focused on home-outdoor exposure to road traffic noise and comprehensively assessed the whole cortex and major subcortical structures and found no significant results suggestive of gross noise-related functional repercussions.However, the subcortical elements of the central auditory pathway were not specifically evaluated in this study.Therefore, a detailed, auditory system-specific analysis may be valuable to potentially capture more precise effects of road traffic noise on functional connectivity during brain development.
While excessive noise levels can directly lead to hearing impairment, noise may also lead to a neuropathy of the inner ear without impacting the hearing threshold, but altering hearing perception, which could affect learning (Plack et al., 2014).Moreover, a previous fMRI study observed that moderate exposure to noise in rats (65 dB, 2 months, i.e. sub-chronic exposure) was able to create changes in the geniculate body and auditory cortex (Lau et al., 2015a).Such effect may be even more relevant during childhood or even later in adolescence, when elements of the central auditory pathway are still developing, as observed in the brainstem function and maturation of obligatory cortical auditory evoked potential components (Skoe et al., 2015;Wunderlich et al., 2006).The understanding of how environmental noise could affect the brain development of the specific structures involved in hearing is also important given that it has been demonstrated in mammalians studies that the auditory pathway is a highly distributed network, which exhibits extensive developmental neuroplasticity, as well as notable adaptation to adverse conditions and disease (Bandyopadhyay et al., 2010;Schreiner and Polley, 2014;Lau et al., 2015a).At the same time, little is yet known about how this neuroplasticity could impact learning (Schreiner and Polley, 2014).Besides, childhood is a crucial period of overall brain maturation (Stansfeld and Clark, 2015;Pujol et al., 2016) and children have been described as a particularly vulnerable population to the effects of environmental noise, particularly for transportation noise at school in relation to cognitive development (Clark and Paunovic, 2018;Clark et al., 2020;Erickson and Newman, 2017;Foraster et al., 2022;van Kamp and Davies, 2013).We thus postulate that prolonged exposure to road traffic noise at school during brain development would affect fine neuronal coupling at the basic levels of auditory signal processing.So, on one hand, persistent auditory input could hypothetically accelerate the maturation of primary neural connections in the auditory pathway.On the other, environmental noise might interfere with other relevant sounds that need processing, creating a low quality auditory stimuli which could ultimately hamper the maturation of high-order connections at the cortex.
The central auditory pathway begins when the cochlear nerve enters the brainstem.The neural activity is firstly processed in several nuclei (e. g., cochlear nuclear complex, superior olive and medial nucleus of trapezoid body) and is then convergently transmitted to the inferior colliculus that is a hub receiving both the auditory input (Ress and Chandrasekaran, 2013) and descending projections from the auditory cortex (Winer, 2005;Stebbings et al., 2014) and primarily processing the basic aspects of the auditory signal (De Martino et al., 2013;Moerel et al., 2015;Ress and Chandrasekaran, 2013).The inferior colliculus projects ipsilaterally and contralaterally to the medial geniculate body, which is the thalamic relay of the pathway that further contributes to encoding the frequency and localization of sound (Mihai et al., 2019).From the thalamus, the auditory signal is transmitted to the primary auditory cortex.
The aim of the present study was to assess the association between school-indoor exposure to annual average road traffic noise and functional connectivity of key elements of the central auditory pathway in schoolchildren.The inferior colliculus, medial geniculate body and primary auditory cortex were thus specifically targeted to generate functional connectivity MRI maps using a region-of-interest approach.

Participants
This study was developed in the context of a large-scale project designed to assess the effects of environmental factors on brain development in children (BREATHE project).Study design and participant selection have been described in full detail elsewhere (Sunyer et al., 2015;Pujol et al., 2016).In short, the BREATHE project recruited children to form a representative sample of schoolchildren in Barcelona (Sunyer et al., 2015).A representative subsample of 278 children from 34 schools was recruited from the larger cohort to participate in neuroimaging.The MRI procedure was explained in detail to the parents or legal guardians by phone, prior to providing the informed consent to participate in the study.All the MRI procedure was explained again by a trained investigator and any clarification was also provided to both children and parents before starting the scan acquisition.There were children that did not want to perform the MRI before it started, and children that wanted to interrupt the acquisition once it had started.In the current study, we included 229 children (49.2% girls) with complete noise exposure assessment and valid fMRI data.The mean age at MRI was 9.7 years, SD 0.8 years and range 8.0-12.1 years.

Noise exposure assessment and contextual characteristics
We measured road traffic noise in two campaigns, the first starting in January 2012 (cold season) and the second 6 months later (warm season).In each campaign, noise was assessed in the classroom at each school for 2 days.We followed the protocol of ISO 1996ISO -2 (2007) ) which allows to obtain long-term noise levels performing short-term noise measurements.We assessed LAeq in decibels (dB), i.e.A-weighted equivalent sound pressure levels, which represents average noise levels.Measurements were performed through three consecutive10-min recordings in different positions in one classroom with a calibrated CESVA SC160 device.Measurements were obtained with closed windows and before lessons started to obtain background road traffic noise levels.This was repeated again the day after.Measurements were supervised to identify irrelevant sounds and these were later removed to procure a specific measure of road traffic noise.Due to incomplete data during the first campaign, the exposure used in the present study was the 2-day average of the second campaign measurements.
The intraclass correlations (ICC) of the means of each campaign were very high (i.e.ICC = 0.89 (between the 2-day mean of each campaign) and ICC = 0.94 (mean of the 2 campaigns).This supports the use of the 2-day average of the second campaign as a marker of long-term schoolindoor exposure to road traffic noise.
Additional characteristics were collected, as previously described (Sunyer et al., 2015), including age and sex of the child, maternal education, home and school vulnerability index at census tract level, behavioral problems (Strengths and Difficulties Questionnaire) answered by parents, and annual average levels of traffic-related air pollution (i.e.nitrogen dioxide and elemental carbon) measured in the classroom during the campaigns of year 2012.

MRI acquisition
MRI was acquired with a 1.5 T Signa Excite system (General Electric, Milwaukee, WI, USA), which was equipped with an eight-channel phased-array head coil and used a single-shot echoplanar imaging (EPI) software.The complete study protocol included a high-resolution 3D anatomical sequence, diffusion tensor imaging (DTI) of 25 directions, proton (1 H) magnetic resonance spectroscopy based on the Proton Brain Exam-Single Voxel (PROBE-SV) software and two functional MRI sequences (task-activation and resting-state functional imaging).The acquisition details of each sequence are provided elsewhere (Pujol et al., 2016).The current study was based on the specific analysis of the functional MRI resting-state sequence that consisted of gradient recalled acquisition in the steady state with repetition time of 2000 ms, echo time of 50 ms; field of view of 24 cm, 64 × 64-pixel matrix, pulse angle of 90 • , slice thickness of 4 mm (inter-slice gap, 1.5 mm).A total of 22 interleaved slices parallel to the anterior-posterior commissure line covering the brain were prescribed.The sequence was taken in resting state and lasted 6 min in total.180 whole-brain EPI volumes were obtained from the scan.To allow magnetization to reach equilibrium, the first four (additional) images in each run were discarded.Children were instructed to relax, keep their eyes closed, and lie still during the scan.

Image preprocessing
Standard procedures previously used by our group were considered for the present functional MRI analysis (Pujol et al., 2016).The steps for image preprocessing consisted of motion correction, spatial normalization and smoothing carried out with a Gaussian filter (full-width half-maximum, 8 mm).The data was directly normalized to the standard Statistical Parametric Mapping (SPM) EPI template and re-sliced to 2 mm isotropic resolution in Montreal Neurological Institute (MNI) space.Low-frequency drifts below 0.008 Hz were removed using a Discrete cosine Transform (DCT) filter.Furthermore, we obtained estimates of cerebrospinal fluid (CSF), white matter, and global brain signal fluctuations (using standard masks in MNI space from SPM), which were included as nuisance (confounding) variables in the regression analyses.
To control for potential head motion effects, the next steps were adopted: a) Conventional alignment of SPM time-series to the first image volume in each subject; b) Exclusion of 22 participants (from an original group of 251 children with complete resting-state fMRI) due to large head motion (boxplot-defined outliers) (Pujol et al., 2014), leading to a final sample of 229 children; c) Inclusion of motion-related regressors and estimates of global brain signal fluctuations as confounding variables in first-level (single-subject) analyses; d) To discard motion-affected volumes within subject, censoring-based MRI signal artifact removal (scrubbing) (Power et al., 2014) was used.For each individual, inter-frame motion measurements (Pujol et al., 2014) were used as an indicator of data quality to flag volumes of questionable quality across the run.For points with inter-frame motion >0.2 mm, the corresponding volume as well as the immediately previous and subsequent two volumes were removed.This led to the exclusion of a mean of 10.8 (6.0%) ± 13.4 vol out of 180 fMRI resting-state sequence volumes; e) Further exclusion of potential motion effects applying a summary measurement for each subject (mean inter-frame motion across the fMRI run) as a regressor in the second-level (group) analyses in SPM (Pujol et al., 2014).

Definition of connectivity maps
The objective of this study was to selectively analyze functional connectivity of key nodes of the central auditory pathway up to the primary auditory cortex (Berlot et al., 2020;Moerel et al., 2015;Schelinski et al., 2022).Thus, we targeted the inferior colliculus, the medial geniculate body of the thalamus and the primary auditory cortex.
Based on previous studies (Díaz et al., 2012;Fitzhugh et al., 2019;Sitek et al., 2019;Schelinski et al., 2022), the regions of interest (ROI) (or 'seeds') were centered at MNI coordinates (in mm): (i) inferior colliculus [x = 5, y = − 35, z = − 10]; (ii) medial geniculate body [x = 16, y = − 24, z = − 6]; and (iii) primary auditory cortex (Heschl's gyrus) [x = 45, y = − 19, z = 7].Each seed region was defined as a 3.5-mm radial sphere using the MarsBar ROI toolbox in MNI stereotaxic space (Brett et al., 2003).Next, the mean ROI value at each time point over the time-series was computed to extract the signals of interest for each seed region.To generate the connectivity maps, the signal time course of a selected seed region was applied as a regressor to be correlated with the signal time course of every voxel in the brain to produce first-level (single-subject) voxel-wise statistical parametric maps.

Association between noise and connectivity maps: voxel-wise correlation analysis
After individual preprocessing of each functional image sequence, separate second-level analysis were performed using SPM to map voxelwise across-subject correlation between school-indoor noise (exposure) and the selected functional connectivity maps (outcome) of the central auditory pathway.A motion summary measure (mean interframe motion [Pujol et al., 2014]) for each participant was included as a nuisance variable.
The results were considered significant at a height threshold of p < 0.005, satisfying the family-wise error (FWE) rate correction PFWE <0.05 using SPM.Considering study aims and the small size of the anatomical structures examined (Moerel et al., 2015;Ress and Chandrasekaran, 2013;Sitek et al., 2019), a small volume correction approach was adopted to test for significant results within the assessed levels of the auditory pathway.Results anatomically related to the inferior colliculus level were corrected for the estimated volume of the brainstem (WFU Pickatlass), those related to the medial geniculate body were corrected for the volume of the thalamus (WFU Pickatlass; Keller et al., 2012), and the cortical level for the volume of the primary auditory cortex (WFU Pickatlass).A whole-brain correction thresholding was considered to test for significant results between each auditory pathway region-of-interest and the rest of the brain.

Sensitivity analyses
The effect of potential confounders on the association between noise and functional connectivity was tested (one-by-one and combined) for each significant finding including age, sex, parental education, home and school vulnerability index, behavioral problems, or annual average levels of traffic-related air pollution using multivariate linear regression models.Similarly, given that the voxel-wise correlation analysis did not allow for multilevel analysis, we tested the effect of school using a linear mixed-effect model between noise and functional connectivity, including all the confounders above plus a random effect by school.

Results
Table 1 summarizes the characteristics of the study sample.The annual average school-indoor noise levels showed a mean of 37.4 dB (SD 4.7 and range 28.8-48.0dB) and median (interquartile range, IQR) of 39.9 (8.4) dB, which exceeded the maximum background noise level (35 dB) recommended in school classrooms (Berglund et al., 1999).Children had a mean (SD) age of 9.7 (0.8) years, median (IQR) of 9.8 (1.3), 49.2% were girls, 67.7% of mothers and 57.8% of partners had university studies.No differences were observed between the included MRI sample and the excluded MRI sample except that the included sample was slightly older (0.1 y) and had slightly less girls (1.9%) G. Martínez-Vilavella et al. compared to the total recruited MRI sample.The annual average school-outdoor noise levels showed a mean of 62.4 (6.4) dB and a Spearman's correlation with school-indoor noise levels of 0.15 (p-value = 0.023).

Functional connectivity maps
The functional connectivity map generated from a region-of-interest (seed) placed at inferior colliculus identified a network that bilaterally included the upper brainstem, the cerebellum, the thalamus-pulvinar, hypothalamus, the lentiform nuclei and several areas of the cerebral cortex such as visual cortex, insula, anterior cingulate cortex and superior temporal gyrus (Table 2).
The medial geniculate body functional connectivity map included the contralateral thalamus, and bilaterally the upper brainstem with both colliculi, the lentiform nuclei, the cerebellum and cortical areas such as supplementary motor cortex, sensory-motor cortex, visual cortex, lateral temporo-occipital, insula, fronto-parietal operculum and superior temporal gyrus.
The primary auditory cortex functional connectivity map identified a network that included the contralateral auditory cortex, and bilateral insula, fronto-parietal operculum, sensory-motor cortex, visual cortex, anterior cingulate cortex, and subcortical areas such as the midbrain, hypothalamus, thalamus and the colliculi in the upper brainstem.
Finally, the functional connectivity maps obtained from our selected ROIs reliably captured the relevant brain structures processing auditory stimuli (Boyen et al., 2014;Smith et al., 2009;Heckner et al., 2021).For example, our extensive primary cortex seed map included 63.8% of the network identified in the meta-analysis by Heckner et al. (2021) based on 122 task-activation auditory experiments.Fig. 1 comparatively illustrates the functional anatomy of both our resting-state primary auditory cortex seed map and the set of regions of significant convergence across tasks during auditory stimuli processing.

Associations between exposure to road traffic noise and functional connectivity
Significant and anatomically specific associations between schoolindoor road traffic noise and functional connectivity measures were observed in the inferior colliculus and medial geniculate body maps.Within the inferior colliculus functional connectivity map, measured noise was associated with higher connectivity between the inferior colliculus seed region and a bilateral thalamic region adjacent to the medial geniculate body involving part of the pulvinar nucleus (Fig. 2 and Table 3).Reciprocally, in the medial geniculate body map, noise was associated with higher connectivity between the medial geniculate body seed and a bilateral region adjacent to the inferior colliculus (Fig. 3).Therefore, school-indoor exposure to road traffic noise in children was associated with stronger connectivity notably within the subcortical auditory pathway (Table 3).No significant findings were observed involving more distant structures at the whole-brain threshold.
As to the primary auditory cortex (Heschl's gyrus) functional connectivity map, no significant results were observed both at whole-brain and auditory-pathway thresholding.However, it is of interest to mention  x y z are coordinates given in Montreal Neurological Institute (MNI) space.
Statistics correspond to a corrected threshold P FWE < 0.05 estimated using SPM.
T denotes statistic transformed of Pearson's r.
that noise showed a negative correlation with the strength of connectivity between the primary auditory cortex seed region and bilateral inferior colliculus in a highly specific manner (Fig. 4).Therefore, the direction of potential noise effects (strengthen versus weaken functional connectivity) was distinct for both cortical and subcortical levels of the auditory pathway.
No confounding effect was observed when including age, sex, parental education, home and school vulnerability index, behavioral problems, annual average levels of traffic-related air pollution, or school as random effect.In specific, decreases in β coefficients after the inclusion of single confounders or their combination were marginal (mean ± SD, 0.9% ± 1.5%) with no variables affecting the primary results with β reductions greater than 5%.

Discussion
To our knowledge, this is the first study to investigate the potential repercussions of road traffic noise in schools on functional connectivity   x y z, coordinates given in Montreal Neurological Institute (MNI) space.Statistics at corrected threshold P FWE < 0.05 using SPM Small volume correction.a High threshold of p < 0.05 not satisfying the SPM small volume correction.
in the central auditory pathway in primary school children.Schoolindoor noise was significantly associated with stronger connectivity between the inferior colliculus and a bilateral thalamic region adjacent to the medial geniculate body, and with stronger connectivity between the medial geniculate body and a bilateral brainstem region adjacent to the inferior colliculus.Such a connectivity strengthening effect did not extend to the cortical level of the auditory pathway (Heschl's gyrus).
Rather, if at all, the direction of such association showed an opposite (weakening) tendency.Despite the different noise exposure assessment, which prevents full comparison, the absence of significant findings in distant brain areas would be in line with the reported lack of any effect of residential exposure to road traffic noise on cortical functional connectivity in the comprehensive imaging assessment by Pérez-Crespo et al. (2022) also in a schoolchildren cohort.Therefore, the data could potentially indicate that the repercussions of exposure to road traffic noise might be subtle  and system-specific, with significant associations limited within the subcortical portion of the central auditory pathway.Remarkably, alterations in the subcortical structures of the auditory pathway due to chronic noise exposure have also been reported in rodents (Chang and Merzenich, 2003;Lau et al., 2015aLau et al., , 2015b)).
We observed a positive association between exposure to noise and the strength of functional connectivity.In the context of development, such a direction of findings can be interpreted as expressing higher within-network integration (Cao et al., 2017;Oldham and Fornito, 2019;Pujol et al., 2016).Excessive noise exposure is believed to place extra sensory demands on children (Simon et al., 2022), and previous studies have demonstrated that background noise in school environments may indeed increase, for instance, listening effort (Peelle, 2018;Prodi and Visentin, 2015).Considering that the maturation of the auditory pathway is not yet complete in school-age children, and the elevated plasticity potential of this pathway (Lau et al., 2015a), a higher sensory demand may plausibly influence the developmental shaping of the pathway (Simon et al., 2022).Our hypothesis is that higher noise exposure, as a form of soft but sustained biological stress, would accelerate the maturation of the central auditory pathway in the form of earlier functional coupling between its basic levels.
Such acceleration of maturation of stem elements in the auditory pathway, however, could interfere with the developmental progress of more distal levels that mature later, leading to a faster, but ultimately less complete, auditory system development (Tooley et al., 2021).In this regard, it is important to indicate that previous mammalian studies have demonstrated that environmental noise may delay the organizational maturation of the auditory cortex (Chang and Merzenich, 2003).In fact, in our study, we found no evidence of a strengthening effect of environmental noise on the connectivity of the primary auditory cortex.Moreover, as emphasized above, the correlation of noise specifically with the connectivity between the primary auditory cortex and the inferior colliculus showed the opposite direction, which could potentially indicate a lower maturation, although this association was not significant.Interestingly, the descending cortical projection of the auditory pathway bypasses the medial geniculate body on its way to the inferior colliculus (Javad et al., 2014;Palmer and Rees, 2010).Further research is warranted to establish whether prolonged noise exposure could ultimately interfere with the maturation of the highest-order processes in the auditory system and whether this could affect attention or learning, as observed in epidemiological studies (Clark et al., 2020;Foraster et al., 2022).
As to the specificity of the anatomical results for the colliculus, noise was associated with higher connectivity between the region of interest (inferior colliculus) and a region adjacent to the medial geniculate body of the thalamus.Both the inferior colliculus and the medial geniculate body are key auditory pathway nodes.However, it was perhaps somewhat surprising that the implicated region included a part of the pulvinar, which is the posterior nuclear complex in the thalamus most frequently associated with processing of visual information (Kurzawski et al., 2022;Purushothaman et al., 2012;Zhou et al., 2016).Nonetheless, previous studies in humans and primates have demonstrated that the pulvinar has a key role in the processing, regulation and integration of multisensory information, including auditory signals (Froesel et al., 2021;Chou et al., 2020).The role of the pulvinar may be particularly relevant in ambiguous (e.g., noisy) environments, where sensory perception relies on the combination of sensory information from various sources (Froesel et al., 2021).
One of the common limitations when using MRI in children and especially in functional MRI, is the potential impact of head movements on the quality of the image.We considered this issue carefully and adopted several means to rigorously control the effects of motion, as indicated in the methods.Also, a higher MRI signal may be obtained using a higher magnetic field (i.e., 3-T magnets or even ultra-high fields at 7 T).Another limitation is the use of a 1.5-T magnet instead of 3 T option.The 3 T option was available, however, the FP7-ERC Ethics Review Committee suggested to limit magnetic field strength in children.Related to the latter, another limitation of the study is the rather low spatial resolution of the MRI images and the smoothing applied in the preprocessing analysis (8 mm FWHM), specially for identifying small structures as, for example, the medial geniculate body.A greater resolution could have improved the precision and strength of the results.Finally, some exposure misclassification in noise levels may be present due to the shift of classrooms between the 2 academic years covered by this study or the position of some classrooms towards the street.As this error may act in both directions, we cannot exclude non-differential misclassification, which could have biased our results towards the null.The main strengths of the study include the large sample size for a neuroimage study and the detailed exposure assessment of road traffic noise, carried out in the classroom, where children spent most of the time and concentration occurs.Moreover, the exposure assessment was representative of long-term exposure during the study period, as observed by the high intraclass correlation between the 2-day measurement campaigns as reported in the methods.We also expect measurements taken before lessons started to be representative of the daily exposure to road traffic noise, given that the road traffic noise levels measured outside the school before lessons started were highly correlated with the annual average noise levels for the daytime (Spearman's rank r = 0.86) estimated at the same position from Barcelona's Strategic Noise Map for year 2012 (European Commission Directive, 2002).
In conclusion, our results have revealed that long-term exposure to road traffic noise inside schools was associated with stronger functional connectivity in the subcortical relays of the auditory pathway in primary school children.This finding may suggest that prolonged noise exposure to road traffic noise in developing individuals may accelerate maturation of basic elements of the central auditory system, which could ultimately interfere with the developing potential in the whole auditory system.

Fig. 1 .
Fig. 1.Render display of the primary auditory cortex resting-state seed map generated in our study (A), task-based auditory network identified by Heckner et al. (2021), meta-analysis (B) and their overlap (C).All the maps are thresholded at cluster-level p < 0.05, family-wise error-corrected (FWE) for multiple comparisons, with a cluster-forming threshold at p < 0.001.The map of the Heckner et al. (2021) meta-analysis was publicly available via the ANIMA database (https://anima.inm7.de).

Fig. 2 .
Fig. 2. Results from the functional connectivity MRI analysis.Representative views of the functional connectivity map generated from a region-of-interest placed at the inferior colliculus are shown in the top images.Bottom images illustrate the regions showing a significant positive association between school-indoor exposure to road traffic noise (LAeq, in dB) and functional connectivity.T denotes statistic transformed of Pearson's r.

Fig. 3 .
Fig. 3. Representative views of the functional connectivity map generated from a region-of-interest placed at the medial geniculate body of the thalamus are shown in the top images.Bottom images illustrate the regions showing a significant positive association between school-indoor exposure to road traffic noise (LAeq, in dB) and functional connectivity.T denotes statistic transformed of Pearson's r.

Fig. 4 .
Fig. 4. Representative views of the functional connectivity map generated from a region-of-interest placed at the primary auditory cortex (top), and the region showing a negative, albeit not significant, association between school-indoor exposure to road traffic noise (LAeq, in dB) and functional connectivity (bottom).T denotes statistic transformed of Pearson's r.

Table 1
Main study sample characteristics and comparison between the included (n = 229) and excluded MRI participants (n = 49).
a Data are median (interquartile range) or percentage (%).bVariables have no missing or less than 10% missing observations.cVariables have no missing or less than 10% missing observations, except LAeq with 55.1% of missing observations and age at MRI with 30.6% of missing observations.dKruskal-Wallis test or Chi-square test for strata of included and excluded participants with continuous or categorical variables, respectively.

Table 2
Functional connectivity maps.

Table 3
Correlations between functional connectivity and school-indoor road traffic noise.