Application of a fussed lasso logistic regression classifier to the study of corpus callosum thickness in early Alzheimer's disease
Babak A Ardekani 1 , Sang Han A Lee 1 , Donghyun Yu 2 , Johan Lim 2 , and Alvin H Bachman 1
The Nathan S. Kline Institute for
Psychiatric Research, Orangeburg, New York, United
Seoul National University, Seoul, Korea
We describe a multi-atlas-based method for corpus
callosum segmentation and a fused Lasso logistic
regression (FLLR) classifier that is able to
differentiate patients with very mild/mild AD from
normal controls (NC) using their CC thickness profile.
We evaluated this technique using data from 196
individuals (98 AD and 98 NC) in the OASIS database. The
FLLR classification accuracy was estimated to be 84%
using cross-validation. Furthermore, the FLLR method
highlights regions of the CC that are significantly
thinner in AD relative to NC. The FLLR model presented
can be extended to include other imaging or chemical
biomarkers of AD.
This abstract and the presentation materials are available to members only;
a login is required.