Comparing FreeSurfer with manual segmentation in the basal ganglia and thalamus of 7 year old children
Wai Yen Loh 1,2 , Zohra M Ahmadzai 2,3 , Lillian Gabra Fam 2,3 , Alan Connelly 1 , Alicia J Spittle 2,4 , Katherine J Lee 2,3 , Terrie E Inder 5 , Jeanie LY Cheong 2,4 , Lex W Doyle 2,4 , Peter J Anderson 2,3 , and Deanne K Thompson 1,2
The Florey Institute of Neuroscience and
Mental Health, Melbourne, VIC, Australia,
Childrens Research Institute, Melbourne, VIC, Australia,
of Paediatrics, University of Melbourne, Melbourne, VIC,
Women's Hospital, Melbourne, VIC, Australia,
and Women's Hospital, Boston, MA, United States
To validate FreeSurfers segmentation method of the
basal ganglia and thalamus in a paediatric population,
FreeSurfers semi-automated segmentations were compared
with manual segmentations in a 7 year-old cohort
comprised of healthy and preterm children. The
intraclass correlation coefficient scores for
consistency were well-above 0.7 for all structures
except the nucleus accumbens and right caudate.
Similarly, Bland-Altman plots indicated that there was a
consistent bias between both methods. This study
suggests that while FreeSurfer is useful in comparing
basal ganglia and thalamus volume differences between
groups, it may not be accurate in obtaining exact
volumes for a paediatric population.
This abstract and the presentation materials are available to members only;
a login is required.