Assumed AIFs in DCE-MRI: Which performs best for assessing breast cancer response?
David K Woolf 1 , N. Jane Taylor 2 , Nina Tunariu 3 , Andreas Makris 1 , Sonia P Li 1 , Mei-Lin W Ah-See 1 , Mark J Beresford 4 , J. James Stirling 2 , James A d'Arcy 3 , David J Collins 3 , and Anwar R Padhani 2
Academic Department of Oncology, Mount
Vernon Hospital, Northwood, Middlesex, United Kingdom,
Strickland Scanner Centre, Mount Vernon Hospital,
Northwood, Middlesex, United Kingdom,
Cancer Imaging Centre, Institute of Cancer Research &
Royal Marsden Hospital, Sutton, Surrey, United Kingdom,
United Hospital Bath, Bath, United Kingdom
We evaluated the performance of six modelled Arterial
Input Functions (AIFs) used with the standard Tofts
model in the setting of primary breast cancer treatment
with neoadjuvant (NAC) chemotherapy. Quantitative
parameters were calculated for 27 patients at baseline
and after 2 cycles of NAC. We assessed the models
abilities to fit patient DCE-MRI data, provide
physiologically-plausible results, and to predict
pathologic complete response (pCR). The modified
Fritz-Hansen model performed best with area under the
ROC curve 0.74 and a sensitivity of 87.5% for pCR
prediction. The Weinmann and femoral artery AIFs
performed least well.
This abstract and the presentation materials are available to members only;
a login is required.