Guenther Grabner1, Katja Pinker2, Stephan Gruber3, Wolfgang Bogner1, Thomas Helbich2, Siegfried Trattnig1,4
1MR Centre of Excellence, Medical University of Vienna, Vienna, Austria; 2Department of Radiology, Medical University of Vienna, Vienna , Austria; 3MR Centre of Excellence, Medical University of Vienna, Vienna , Austria; 4Department of Radiology, Medical University of Vienna, Vienna, Austria
This work is focused on automatic classification of lesions as malignant or benign and on the bottleneck of DCE-MRI, the trade-off between spatial and temporal resolution. The DCE-MRI approach presented here combines high spatial and temporal resolution by splitting the DCE-MRI imaging protocol into three parts (high temporal resolution measurements; a high spatial resolution measurement at contrast maximum followed by high temporal resolution measurements). Enhancement curves obtained from 3D manually drawn regions of interest were fitted using a modified asymmetric logistic model. Model and secondary parameters were successfully classified as malignant or benign using k-means.