Abstract #4090
MR Estimation of Permeability Parameters in Dynamic Contrast Enhanced Studies Using Model Averaging Technique and Nested Model Selection Method
Hassan Bagher-Ebadian 1,2 , Siamak P. Nejad-Davarani 3,4 , James R Ewing 2,3 , Tom Mikkelsen 5 , Rajan Jain 6 , Lisa Scarpace 5 , and Hamid Soltanian-Zadeh 1,7
1
Radiology, Henry Ford Hospital, Detroit, MI,
United States,
2
Physics,
Oakland University, Rochester, MI, United States,
3
Neurology,
Henry Ford Hospital, Detroit, MI, United States,
4
Biomedical
Engineering, University of Michigan, Ann Arbor, MI,
United States,
5
Neurosurgery,
Henry Ford Hospital, Detroit, MI, United States,
6
Radiology,
NYU Langone Medical Center, NY, United States,
7
CIPCE,
ECE Dept., University of Tehran, Tehran, Iran
A nested model selection (NMS) technique along with
physiological concepts of the models is introduced and a
model-averaging technique in Dynamic-Contrast-Enhanced
(DCE)-MR model selection using the Akaike-Information-Criterion
(AIC) is constructed. The Models in NMS are recruited in
the AIC and applied to an exemplary DCE-MR data of a
patient with Glioblastoma-Multiforme. Model-choice and
probability maps estimated from both techniques are
compared. The AIC and NMS provide unique set of
probability maps for estimating the contribution of each
model in a specific voxel. These probabilities allow
combining the estimations from different models, thus
generating a more accurate estimate of permeability
parameters.
This abstract and the presentation materials are available to members only;
a login is required.
Join Here
