Brain network approaches to characterise structure-function relationships after childhood head injury

Abstract

Traumatic brain injury (TBI) is a common leading cause of disability for children and young adults. Such an insult during development leads to a cognitive-behavioural syndrome of impairments post-injury however, the trajectory and magnitude of this syndrome at an individual-patient level are unknown. Pathological effects of a TBI on neuroanatomy can be quasi-quantitatively measured using magnetic resonance imaging (MRI). The current thesis highlights the current state of research into neuroanatomical differences post-TBI and identifies previous investigations of neuroanatomical correlates of later functioning adopt mostly reductionist and univariate approaches. It was posited that utilising MRI methodologies that respect the complex, interrelated nature of neuroanatomy across the cortex would provide better understanding of the neuroanatomical correlates of later cognitive functioning post-injury. The current thesis investigates novel network-level analyses of neuroanatomy, specifically structural covariance and morphometric similarity approaches, and its relationship with neuropsychological functioning, with a focus on executive functioning (EF) at 2 years post-injury. The data used included 107 survivors of paediatric TBI and 36 typically developing controls. Using a structural covariance methodology the current thesis provides evidence that executive dysfunction is associated with atrophic neuroanatomical-changes to topologically important brain-regions within the network. Results also provided evidence that the magnitude of EF difficulties was associated with the extent to which an injury diverted the brain from the ‘typical’ organization of the neuroanatomical network. Using individual morphometric similarity approaches coupled with supervised machine learning, a pattern of morphology, centred in the pre-frontal cortex, predicted later EF. Overall, these findings fit with the hypothesis that network-level neuroanatomical correlates of EF would be found. The current thesis concludes that, regarding the neuropsychological sequalae post-neurological insult, quantification of the complex organisation of neuroanatomy across the cortex is a useful biomarker. Future investigations integrating neuropsychology and neuroimaging to understand brain structure-function relationships should continue to utilise modern network approaches which capture the diffuse, disconnecting nature of injury.