Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.650710
Title: Automatic MRI segmentation of the developing neonatal brain
Author: Makropoulos, Antonios
ISNI:       0000 0004 5357 1157
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2014
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Abstract:
Detailed morphometric analysis of the neonatal brain is required to characterise normal brain development and investigate the neuroanatomical correlates of cognitive impairments. The segmentation of the brain in Magnetic Resonance Imaging (MRI) is a prerequisite to obtain quantitative measurements of regional brain structures. These measurements obtained at term-equivalent or early preterm age may lead to improved understanding of brain growth and may help evaluate long-term neurodevelopmental performance at an early stage. This thesis focuses on the development of an accurate segmentation algorithm for the neonatal brain MR images and its application in large cohorts of subjects. Neonatal brain segmentation is challenging due to the large anatomical variability as a result of the rapid brain development in the neonatal period. The lack of training data in the neonatal period, encoded in brain atlases, further hinders the development of automatic segmentation tools. A novel algorithm for the tissue segmentation of the neonatal brain is proposed. The algorithm is extended for the regional brain segmentation. This is the first segmentation method for the parcellation of the developing neonatal brain into multiple structures. A novel method is further proposed for the group-wise segmentation of the data that utilizes unlabelled data to complement the labelling information of brain atlases. Previous studies in the literature tended to overestimate the extent of the cortical region. A method based on the morphology of the cortex is introduced to correct for this over-segmentation. The segmentation method is applied on an extensive database of neonatal MR images. Regional volumetric, surface and diffusion tensor imaging measurements are derived from the early preterm period to term-equivalent age. These measurements allow characterisation of the regional brain development and the investigation of correlations with clinical factors. Finally, a spatio-temporal structural atlas is constructed for multiple regions of the neonatal brain.
Supervisor: Rueckert, Daniel; Counsell, Serena; Edwards, David Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.650710  DOI: Not available
Share: