Abstract #0921
Variations in DCE-MRI Assessment of Breast Cancer Therapy Response: A Multicenter Data Analysis Challenge
Wei Huang 1 , Xin Li 1 , Xia Li 2 , Ming-Ching Chang 3 , Matthew J Oborski 4 , Dariya I Malyarenko 5 , Mark Muzi 6 , Guido H Jajamovich 7 , Andriy Fedorov 8 , Yiyi Chen 1 , Alina Tudorica 1 , Sandeep N Gupta 3 , Charles M Laymon 4 , Kenneth I Marro 6 , Hadrien A Dyvorne 7 , James V Miller 3 , Thomas L Chenevert 5 , Thomas E Yankeelov 2 , James M Mountz 4 , Paul E Kinahan 6 , Ron Kikinis 8 , Bachir Taouli 7 , Fiona Fennessy 8 , and Jayashree Kalpathy-Cramer 9
1
Oregon Health & Science University,
Portland, Oregon, United States,
2
Vanderbilt
University, Nashville, Tennessee, United States,
3
General
Electric Global Research, Niskayuna, New York, United
States,
4
University
of Pittsburgh, Pittsburgh, Pennsylvania, United States,
5
University
of Michigan, Ann Arbor, Michigan, United States,
6
University
of Washington, Seattle, Washington, United States,
7
Icahn
School of Medicine at Mount Sinai, New York, New York,
United States,
8
Brigham
and Womens Hospital and Harvard Medical School, Boston,
Massachusetts, United States,
9
Massachusetts
General Hospital and Harvard Medical School, Boston,
Massachusetts, United States
Seven institutions of the NCI-sponsored Quantitative
Imaging Network (QIN) participated in a DCE-MRI data
analysis challenge, in which 12 pharmacokinetic
models/algorithms (including Tofts model, extended Tofts
model, and Shutter-Speed model) were used to analyze
shared breast DCE-MRI data collected at one center
before and after one cycle of neoadjuvant chemotherapy,
from 10 breast cancer patients. Tumor ROI definition,
AIF, and precontrast T1 were fixed for analysis of each
data set across all algorithms. Considerable variations
in DCE-MRI parameters were found among the algoritms
with Ktrans wCV as high as 0.59. Encouragingly, Ktrans
and kep values after one therapy cyles and their %
changes (relative to baselin) obtained from all
algorithms provided good to excellent early prediction
of pathologic response.
This abstract and the presentation materials are available to members only;
a login is required.
Join Here
