Abstract #3503
Robustness of a fully automated brain segmentation tool for multiple MRI protocols: test for clinical applications
Zifei LIANG 1,2 , Xiaohai HE 1 , Andreia V. Faria 2 , Kenishi Oishi 2 , Yue Li 3 , Kinya Okada 2,4 , Can Ceritoglu 5 , Xiaoying Tang 5 , Michael Miller 5 , and Susumu Mori 2,6
1
College of Electronics and Information
Engineering, Sichuan University, Chengdu, Sichuan,
China,
2
Johns
Hopkins University School of Medicine, BALTIMORE, MD,
United States,
3
AnatomyWorks,LLC,
BALTIMORE, MD, United States,
4
MitsubishiTanabe
Pharma Corporation, Kawagishi, Japan,
5
Center
for Imaging Science, Johns Hopkins University,
BALTIMORE, MD, United States,
6
Kennedy
Krieger Institute, BALTIMORE, MD, United States
We tested the robustness of a multi-atlas whole-brain
segmentation tool against different imaging protocols.
We measured the volumes of 286 structures in 72 healthy
brains from ADNI database, from three scanner
manufacturers and two field strengths. The protocol
impact, that explained 1.5% of the data variation, is
far smaller than age effect, that explained 10.4% of the
data variation, indicating that the data pooled from
multiple sources can be used to evaluate biological
effects. This type of robust technology is a key to
apply quantitative analysis for clinical diagnosis, in
which highly consistent image protocol cannot be
expected.
This abstract and the presentation materials are available to members only;
a login is required.
Join Here
