Comparison of segmentation techniques to measure tissue-specific atrophy in Multiple Sclerosis
Patricia Alves Da Mota 1 , Ferran Prados 2 , Wallace J Brownlee 1 , Manuel Jorge Cardoso 2 , Matteo Pardini 1 , Nicolas Toussaint 2 , Declan T Chard 1 , Sbastien Ourselin 2 , David H Miller 1 , and Claudia AM Wheeler-Kingshott 1
NMR Research Unit, Department of
Neuroinflammation, Queen Square MS Centre, UCL Institute
of Neurology, London, England, United Kingdom,
of Medical Physics and Bioengineering Wolfson House,
Translational Imaging Group CMIC, London, England,
Brain atrophy is a well established pathologically
feature of multiple sclerosis (MS). Here we compare GIF,
a novel segmentation technique which has not been
previously applied in people with MS, to SPM12 results.
Acknowledging the small number of subjects, statistical
measures and qualitative assessment showed that GIF
provides a relatively significant improvement in
segmentation accuracy when compared to SPM12, with an
improved agreement between observers (r=0.740 VS
r=0.203). GIF was also found to be more robust (i.e.
lower standard deviations) at estimating brain atrophy
(i.e. volume fractions) on both HC and MS subjects.
This abstract and the presentation materials are available to members only;
a login is required.