Julia Catarina Krcher1, Volker Johann Schmid1
1Department of Statistics, Ludwig-Maximilians-University, Munich, Germany
Standard one compartment models used for the quantitative analysis of concentration time series in DCE-MRI fail in modeling heterogeneity. We propose a model with two tissue compartments that accounts for heterogeneity within voxels and thus more appropriately describes the uptake behavior at tumor margins. We propose a model selection criterion that accounts for the adequacy both of model fit and of model complexity. DCE-MRI scans from breast cancer data are evaluated with the proposed model selection criterion per voxel: at tumor margins the proposed two tissue compartment model outperforms the standard one compartment model.