Poster #87 - Estimating Hemispheric Specialization in Children, Adolescents, and Emerging Adults

• In Event: Poster Session 19

Abstract

Hemispheric and network specialization appear to reflect the development of large-scale brain architecture, both functional and structural (Hervé et al., 2013). In pre-term infants, there are strong bilateral interhemispheric connections and the emergence of a proto-default network (Fransson et al., 2007). In term infants, network hubs are found in the primary sensory systems, in direct contrast with adult network hubs which are found in the higher association cortices (Fransson et al., 2011). Over infancy, proto-networks gradually become more focused and functional connections between distant intrahemispheric regions begin to appear (Smyser et al., 2010). In early childhood, network asymmetries emerge, with leftward asymmetries of core language regions, such as the inferior frontal gyrus, increasing between 2-7.5 years (Reynolds et al., 2019). As hemispheric specialization increases across childhood, there is an accompanying improvement in visuospatial attention (Everts et al., 2009) and language abilities (Everts et al., 2009; Friederici et al., 2011; Karunanayaka et al., 2010; Ressel et al., 2008). This functional lateralization only increases across adolescence and into adulthood (Friederici et al., 2011; Karunanayaka et al., 2010) and reflects a shift in connectivity from inter-hemispheric to predominantly left hemisphere intra-hemispheric (Friederici et al., 2011). However, it is unknown how hemispheric specialization is characterized across childhood, adolescence, and into adulthood when an individual-level approach is applied.

It is hypothesized that the functional asymmetry for the left- and right-lateralized networks will only be slightly lateralized in childhood yet will increase in lateralization throughout development. To test this hypothesis, a cross-sectional dataset of resting-state functional MRI scans from 652 children and adolescents ages 5-21 years will be used (Somerville et al., 2018). The processing pipeline will include preprocessing, network parcellation, functional connectivity calculations, and the quantification of specialization. For preprocessing, Freesurfer and the CBIG2016 pipeline will be utilized (Kong et al., 2019; Li et al., 2019). This will be following by multi-session hierarchical Bayesian modeling network parcellation (Kong et al., 2019). In a separate stream of processing, functional connectivity matrices will be calculated for each rs-fMRI run and then averaged across runs within an individual using MATLAB R2018b. Specialization will then be calculated on these averaged functional connectivity matrices and is operationalized as the autonomy index (AI; Wang et al., 2014). The AI is calculated as the difference between normalized within- and cross-hemisphere connectivity as follows: AI = Ni/Hi – Nc/Hc where Ni and Nc are the number of vertices correlated to the seed ROI (using a threshold of |0.25|) in the ipsilateral hemisphere and contralateral hemisphere, respectively. Hi and Hc are the total number of vertices in the ipsilateral and contralateral hemisphere, respectively. To compute the specialization of each functional network, the AI will be averaged within the boundary of each network on an individual basis. Multiple regressions in R 4.0.2 will be used to examine network differences in specialization as a function of mean-centered age, in addition to the following covariates: sex, scan site, and mean-centered age.

Authors

• Madeline Peterson, Brigham Young University
  Presenting Author

• Jared Nielsen