Quantification of cortical folding using MR image data

The cerebral cortex is a thin layer of tissue lining the brain where neural circuits perform important high level functions including sensory perception, motor control and language processing. In the third trimester the fetal cortex folds rapidly from a smooth sheet into a highly convoluted arrangement of gyri and sulci. Premature birth is a high risk factor for poor neurodevelopmental outcome and has been associated with abnormal cortical development, however the nature of the disruption to developmental processes is not fully understood. Recent developments in magnetic resonance imaging have allowed the acquisition of high quality brain images of preterms and also fetuses in-utero. The aim of this thesis is to develop techniques which quantify folding from these images in order to better understand cortical development in these two populations. A framework is presented that quantifies global and regional folding using curvature-based measures. This methodology was applied to fetuses over a wide gestational age range (21.7 to 38.9 weeks) for a large number of subjects (N = 80) extending our understanding of how the cortex folds through this critical developmental period. The changing relationship between the folding measures and gestational age was modelled with a Gompertz function which allowed an accurate prediction of physiological age. A spectral-based method is outlined for constructing a spatio-temporal surface atlas (a sequence of mean cortical surface meshes for weekly intervals). A key advantage of this method is the ability to do group-wise atlasing without bias to the anatomy of an initial reference subject. Mean surface templates were constructed for both fetuses and preterms allowing a preliminary comparison of mean cortical shape over the postmenstrual age range 28-36 weeks. Displacement patterns were revealed which intensified with increasing prematurity, however more work is needed to evaluate the reliability of these findings.Open Acces