Loading…
CNS*2020 Online has ended
Welcome to the Sched instance for CNS*2020 Online! Please read the instruction document on detailed information on CNS*2020.
Back To Schedule
Monday, July 20 • 5:40pm - 6:00pm
O12: A Spatial Developmental Generative Model of Human Brain Structural Connectivity

Log in to save this to your schedule, view media, leave feedback and see who's attending!

Feedback form is now closed.


Stuart Oldham, Ben Fulcher, Kevin Aquino, Aurina Arnatkevičiūtė, Rosita Shishegar, Alex Fornito

The human connectome has a complex topology that is thought to enable adaptive function and behaviour. Yet the mechanisms leading to the emergence of this topology are unknown. Generative models can shed light on this question, by growing networks in silico according to specific wiring rules and comparing properties of model-generated networks to those observed in empirical data [1]. Models involving trade-offs between the metabolic cost and functional value of a connection can reproduce topological features of human brain networks at a statistical level, but are less successful in replicating how certain properties, most notably hubs, are spatially embedded [2,3]. A potential reason for this limited predictive ability is that current models assume a fixed geometry based on the adult brain, ignoring the major changes in shape and size that occur early in development, when connections form.

To address this limitation, we developed a generative model that accounts for developmental changes in brain geometry, informed by structural MRIs obtained from a public database of foetal scans acquired from 21–38 weeks gestational age [4]. We manually segmented the cortical surface of each brain and registered each surface to an adult template surface using Multimodal Surface Matching [5,6]. This procedure allowed us to map nodes to consistent spatial locations through development and measure how distances between nodes (a proxy for connectome wiring cost) change through development. We evaluated the performance of classic trade-off models [2] that either assume a fixed, adult brain geometry (static), or those where cost-value trade-offs dynamically change in accordance with developmental variations in brain shape and size (growth). We used connectomes generated from 100 healthy adults with diffusion MRI to benchmark model performance. Model fit was calculated by comparing model and empirical distributions of topological properties. An optimisation procedure was used to find the optimal parameters and best-fitting models for each individual adult brain network [2]. For fair comparison of model fit across models of varying parametric complexity, we used a leave-one out cross- validation procedure.

Spatial models (sptl; which include only distance information) produced poorer fits than those involving distance–topology trade-offs. Homophily models (matching , neighbors; where connections form between nodes with common neighbours) were among the best fitting. Growth models produced slightly better fits than static models overall. These results still generally held when the cross-validation procedure was employed (Fig. 1A). Neither growth nor static models reproduced the spatial topography of network hubs, but growth models are associated with a less centralized anatomical distribution of hubs across the brain, which is more consistent with the empirical data (Fig. 1B).

In summary, we introduce a new framework for examining how developmental changes in brain geometry influence brain connectivity. Our results suggest that while such changes influence network topology, they are insufficient to explain how complex connectivity patterns emerge in brain networks.

References: **[1]** Betzel R, Bassett D. _J. R. Soc. Interface_ 2017; 14. **[2]** Betzel R et al. _NeuroImage_ 2016; __ 124: 1054-64. **[3]** Zhang X et al. _BioRxiv_ 2019. **[4]** Gholipour A et al. _Sci Rep_ 2017; 7. **[5]** Robinson E et al (2014). _NeuroImage_ 100: 414-26. **[6]** Robinson E et al. _NeuroImage_ 2018; 167: 453-65.

Speakers
avatar for Stuart Oldham

Stuart Oldham

Monash University


Monday July 20, 2020 5:40pm - 6:00pm CEST
Crowdcast
  Oral, Structural and Functional Connectivity
  • Moderator Sacha van Albada; Ingo Bojak