Meeting Banner
Abstract #2497

Model Selection in High Temporal Resolution DCE-MRI of Breast Tumors

Robert Luypaert1, Smitha Makkat, Johan De Mey, Steven Sourbron2

1UZ Brussel - Radiology/BEFY, Vrije Universiteit Brussel, Brussels, BE, Belgium; 2Josef Lissner Laboratory for Biomedical Imaging, Ludwig-Maximilian-University, Munich, Germany


Assessment of tumor blood flow and permeability is important for breast cancer treatment. High temporal resolution T1-DCE-MRI and deconvolution provide model independent blood flow values. Extracting permeability information needs model-based analysis. 21 histologically proven breast tumors were examined. The inversion-prepared TFE time courses covered 140s at a temporal resolution of 0.3s. ROI data were fitted to the 2-compartment exchange, modified Tofts and uptake models. The Akaike information criterion was used to investigate which model fitted the data best. Result: in 70% of cases one model was at least twice as likely as the next to be the best.