Tracking and predicting cognitive development using magnetic resonance imaging
Abstract: Neuroimaging of the developing brain has helped describing and quantifying many of the biological processes underlying cognitive development. The protracted development of higher order cognitive functions has allowed detailed description of their neural correlates. While primary sensory and motor functions have been found to be relatively localized, higher order cognitive functions including Working Memory (WM) have been found to be distributed over many brain regions. A growing amount of literature is describing a complex interaction of anatomically separate nodes making up networks sub-serving WM. The development of these networks is dependent on both predetermined maturation and environmental stimulation. The current thesis aims to expand the current knowledge by exploring if WM development can be predicted by using Magnetic Resonance Imaging (MRI) data explaining future development rather than correlating to current capacity. We further apply this principle on a sample of premature born children to predict future cognitive outcome using MRI at birth. Finally we address the question if individual variability in developmental timing affects cognitive abilities in childhood and adolescence. Study I: In this study we show that WM development to some degree can be predicted using structural and functional MRI. The prediction was based on a multivariate model of MRI data and could significantly predict WM two years after the scans. This significance was retained after controlling for three concurrent WM tests. Analysis to localize the predictive effect of MRI suggests basal ganglia and thalamic structures as important for future development while classical cortical WM areas correlate to concurrent WM capacity. Study II: We apply a similar analysis strategy as in Study I on a longitudinal sample of preterm born children. T2 and Diffusion Tensor Imaging (DTI) sequences were collected in the perinatal period and used to predict WM and Numerical Ability (NA) at five and seven years of age. We show that multivariate models can predict NA and WM capacity at five years of age. This was the strongest predictor when compared with previously known important clinical features. T2 based volumetric analysis points towards reductions in insula and basal ganglia volume in the perinatal period among children with low cognitive function at five years of age. Study III: The study explores weather the individual time course of development affects WM abilities when children start school. We trained a multivariate model of brain development using DTI from a sample of normally developing children. We then apply the model on a sample of seven year old children to show that brain maturation correlates strongly with WM abilities while age does not. In summary the articles add to the developmental neuroscience literature by showing the ability of MRI to predict cognitive development. Prediction of development is an area discussed as a promising target for clinical implementation of cognitive neuroimaging. We show the feasibility and clinically relevant effects of prediction in a clinical sample. Finally the measuring of variability in developmental timing in Study III highlight the view of WM development as result of multiple processes.
This dissertation MIGHT be available in PDF-format. Check this page to see if it is available for download.