Mapping anatomical connectivity: a Structural Network Analysis in Early and Profoundly Deaf people.
Francesca Saviola1, Lisa Novello1, Chiara Maffei2, Stefania Benetti1, Ceren Battal3, Stefania Mattioni3, Olivier Marie Claire Collignon1,3, and Jorge Jovicich1

1CIMeC, Center for Mind/Brain Sciences, University of Trento, Rovereto (TN), Italy, 2Athinoula A. Martinos Center, Massachusetts General Hospital and Harvard Medical School, Charlestown, MA, United States, 3Institute of Research in Psychology (IPSY) and in Neuroscience (IoNS), University of Louvain, Louvain-la-Neuve, Belgium


In case of early acquired deafness, auditory deprived temporal regions massively enhance their response to stimuli from remaining senses. This so called cross-modal plasticity also alters functional connectivity between reorganized temporal regions and those from preserved senses. The extent and distribution of white matter structural alterations supporting these functional effects are still poorly understood. In this diffusion MRI study, we investigate white matter reorganization of early deaf relative to hearing controls. Further, since early deaf typically become fluent at sign language, which may itself also induce brain structural reorganization unrelated to deafness, we also include a group of hearing signers.


Deafness is usually accompanied by functional brain alterations that may be thought as an alteration to connectome scaffolding1,2,3,4,5,6. The general goal of this study was to investigate the brain structural network organization in subjects who are early and profoundly deaf. The specific goal was to apply the structural white matter connectome formalism to evaluate network differences in primary, secondary sensory cortices and higher cognitive system regions of deaf subjects relative to healthy controls or hearing signers.


A total of 44 subjects participated in this diffusion MRI study (4T Medspec Bruker): 14 early and profound deaf subjects (ED), 15 age and gender matched hearing controls (HC), and 15 age and gender matched hearing signers (HS). The full-brain diffusion-weighted protocol consisted of 60 diffusion-encoding gradient directions, a b-value of 1500 s/mm2, 2 mm3 isotropic voxels, 0.30 mm axial slice gap,10 volumes without any diffusion-weighting (b0-images),TE = 99 ms and TR = 7100 ms. Network analyses were performed starting from the whole brain unfiltered tractograms, computed using probabilistic spherical-deconvolution-based Anatomically-Constrained Tractography (ACT)7. In each subject’s space, brain structural networks were obtained combining the whole-brain tractography with the anatomical parcellation based on a template of 638 similarly-sized regions built from subdivision of the regions from the Anatomical Automated Labelling Atlas (AAL)8,9,10. After filtering the tractograms11 and co-registering the data in a common subjects’ space (Figure 1), three connectomes were derived for each subject based on the following white matter edge metrics between each pair of parcellation nodes: sum of streamlines weighted for sum of SIFT211 track weights, mean length of interconnecting tracks, mean fractional anisotropy (FA) along tracks. Connectomes were then analysed using Graph Analysis and Network-Based Statistics9, to explore structural connectivity characteristics of deafness.


Network-based comparisons across groups found significant reductions in structural connectivity of ED relative to both HC (Figure 2) and HS (Figure 3) with mean Fractional Anisotropy (p-valueHC-E=0.003; p-valueHS-ED<0.001), mean length of interconnecting tracts (p-valueHC-ED=0.002; p-valueHS-ED=0.003) and number of streamlines (p-valueHC-ED=0.001; p-valueHS-ED=0.001) as metrics. The global network efficiency, defined as the average of inverse shortest path length, did not differ across groups for the FA connectomes but was significantly reduced in ED compared to HC in both the mean tract length (t- valueED-HC=-2.45; p- valueED-HC=0.02) and number of streamlines (t-valueED-HC=-2.11; p- valueED-HC=0.04) connectomes. Node degree, defined as the number of edges connected to a node, was significantly reduced in ED regardless of group (HS and HC) and regardless of connectome (FA, mean tract length, # streamlines), particularly in frontal, parietal and motor regions. Nodal strength, defined as the sum of the weights in edges to the node, showed significant reductions in ED with respect to HC (mainly in frontal, parietal, temporal and motor regions) as well as reductions in ED relative to HS (mainly in parieto-temporal regions). No statistically significant differences were detected, either at network level or in graph metrics, between HC and HS. Overall, ED showed the significant reductions of all metrics (FA, mean tract distance and # of streamlines) in the left hemisphere, both relative to HS and HC


Our study showed, with a variety of diffusion metrics, both global and local reductions in white matter connectivity in the early deaf group compared to hearing signers and healthy controls, previously shown only in grey matter12. Moreover, in line with current evidence13,14,15,16, we detected significant connectivity reductions in the deaf group on the left-hemisphere relative to both hearing controls and hearing signers. Results based on number of streamlines appeared to be more specific and local with respect to FA and mean tract length. The lack of significant differences between HC and HS suggests that white matter alterations resulting from deafness cannot be solely explained by the use of sign language. Therefore, the question of whether hemispheric asymmetry takes place in deafness due to plastic mechanisms or due to sensory deprivation combined with a modified language experience is still an open issue17.


To the best of our knowledge, we identify for the first time both global and local white matter structural connectivity alterations related to deafness, using both network-level characteristics and graph indices. The absence of significant alterations in HS suggests that those effect are linked to auditory deprivation. Further studies are needed to better understand if the observed left hemispheric asymmetry effects in deafness may be related to language skills in the deaf population. In the future, our results and methods may potentially serve the prognostic of the success of cochlear implants18.




  1. Neville H., & Bavelier, D. (2002). Human brain plasticity: evidence from sensory deprivation and altered language experience. Prog Brain Res, 138, 177-88. https://doi.org/10.1016/S0079-6123(02)38078-6
  2. Pavani, F., & Röder, B. (2012). Cross-Modal Plasticity as a Consequence of Sensory Loss: Insights from Blindness and Deafness. In: B. E. Stein (Ed.) “The new handbook of multisensory processing” (pp. 737{759). Cambridge MA: MIT Press.
  3. MacSweeney M., Capek M. C., Campbell R., Woll B. (2008). The signing brain: the neurobiology of sign language. Trends in Cognitive Sciences, 12 (11), 432-440. https://doi.org/10.1016/j.tics.2008.07.010.
  4. Benetti, S., van Ackeren, M. J., Rabini, G., Zonca, J., Foa, V., Baruffaldi, F., … Collignon, O. (2017). Functional selectivity for face processing in the temporal voice area of early deaf individuals. Proceedings of the National Academy of Sciences, 114(31), E6437 LP-E6446. http://www.pnas.org/content/114/31/E6437.abstract
  5. Benetti, S., Novello, L., Maffei, C., Rabini, G., Jovicich, J., & Collignon, O. (2018). White matter connectivity between occipital and temporal regions involved in face and voice processing in hearing and early deaf individuals. NeuroImage, 179(May), 263–274. https://doi.org/10.1016/j.neuroimage.2018.06.044
  6. Kral A., Kronenberger W.G., Pisoni D.B., O'Donoghue G. M.,(2016), Neurocognitive factors in sensory restoration of early deafness: a connectome model, The Lancet Neurology, 15(6), 610-621. https://doi.org/10.1016/S1474-4422(16)00034-X.
  7. Smith, R. E., Tournier, J., Calamante, F., & Connelly, A. (2012). NeuroImage Anatomically-constrained tractography : Improved diffusion MRI streamlines tractography through effective use of anatomical information. NeuroImage, 62(3), 1924–1938. https://doi.org/10.1016/j.neuroimage.2012.06.005
  8. Tzourio-Mazoyer, N., Landeau, B., Papathanassiou, D., Crivello, F., Etard, O., Delcroix, N., et al. (2002). Automated anatomical labeling of activations in spm using a macroscopic anatomical parcellation of the mni mri single-subject brain. Neuroimage 15, 273–289. https://doi.org/10.1006/nimg.2001.0978
  9. Zalesky, A., Fornito, A., & Bullmore, E. T. (2010). NeuroImage Network-based statistic: Identifying differences in brain networks. NeuroImage, 53(4), 1197–1207. https://doi.org/10.1016/j.neuroimage.2010.06.041
  10. 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), 2382–2395. https://doi.org/10.1093/brain/awu132
  11. Smith, R. E., Tournier, J., Calamante, F., & Connelly, A. (2015). NeuroImage SIFT2: Enabling dense quantitative assessment of brain white matter connectivity using streamlines tractography. NeuroImage, 119, 338–351. https://doi.org/10.1016/j.neuroimage.2015.06.092
  12. Li, W., Li, J., Wang, Z., Li, Y., Liu, Z., Yan, F., Xian, J., … He, H. (2015). Grey matter connectivity within and between auditory, language and visual systems in prelingually deaf adolescents. Restorative neurology and neuroscience, 33(3), 279-90. https://doi.org/10.3233/2FRNN-140437
  13. Hribar M., Araujo, A., Battelino, S., & Vovk, A. (2014). Structural alterations of brain grey and white matter in early deaf adults. Hearing Research, 318, 1–10. https://doi.org/10.1016/j.heares.2014.09.008
  14. Kim, D., Park, S., Kim, J., Ha, D., & Park, H. (2009). Alterations of white matter diffusion anisotropy in early deafness. Neuroreport, 20(11), 1032–1036. https://doi.org/10.1097/WNR.0b013e32832e0cdd
  15. Miao W., Li J., Tang M. et al.,(2003). Altered white matter integrity in adolescents with prelingual deafness: a high-resolution tract-based spatial statistics imaging study. American Journal of Neuroradiology, 34(6), 1264–1270. https://doi.org/10.3174/ajnr.A3370
  16. Li J., Li W., Xian J., Li Y., Liu Z., Liu S., Wang X., Wang Z., He H., (2012), Cortical thickness analysis and optimized voxel-based morphometry in children and adolescents with prelingually profound sensorineural hearing loss, Brain Res., 1430, pp. 35-42. https://doi.org/10.1016/j.brainres.2011.09.057.
  17. Karns, C. M., Stevens, C., Dow, M. W., Schorr, E. M., & Neville, H. J. (2017). Atypical white-matter microstructure in congenitally deaf adults: A region of interest and tractography study using diffusion-tensor imaging, Hear Res, 343, 72–82. https://doi.org/10.1016/j.heares.2016.07.008
  18. Heimler, B., Weisz, N., & Collignon, O. (2014). REVIEW REVISITING THE ADAPTIVE AND MALADAPTIVE EFFECTS OF CROSSMODAL PLASTICITY. Neuroscience, 283, 44–63. https://doi.org/10.1016/j.neuroscience.2014.08.003
  19. Avants BB, Tustison NJ, Song G, Cook PA, Klein A, Gee JC. (2011). A reproducible evaluation of ANTs similarity metric performance in brain image registration. NeuroImage,4(3), 2033-44. https://doi.org/10.1016/j.neuroimage.2010.09.025


Figure 1: Image co-registration pipeline. Schematic representation of co-registration steps performed in Advanced Normalization Tools19 on our dataset.

Figure 2: Structural connectivity reorganization in early deaf (ED) relative to healthy controls (HC). Surface rendering of network with significant reduced FA (A), reduced interconnecting tracts (B) and reduced number of streamlines (C and D) ED compared to HC. Network Based Statistics (NBS) detected two significantly different networks in the contrast HC>ED for number of streamlines (panel C and D). NBS permutation testing was thresholded of t-value>3.1, number of permutation 5000, p-value<0.05. Only edges with connections are shown in each panel.

Figure 3: Structural connectivity reorganization in early deaf (ED) relative to hearing signers (HS). Surface rendering of network with significant reduced FA (A), interconnecting tracts (B) and number of streamlines (C) in ED compared to HS detected by the NBS algorithm in the Matlab toolbox. NBS permutation testing was thresholded of t-value>3.1, number of permutation 5000, p-value<0.05. Only edges with connections are shown in each panel.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)