Perfusion Model Approximation in Cerebral Tumours: Feature Images and Computation Times

Ultrasound contrast imaging can be used to visualize tumours due to changes in the local perfusion. Quantitative analysis of the data may establish features usable for classification of the scanned tissue. In this work we approximate model functions for the contrast agent concentration time course of every pixel, while reducing computation times by means of parallel computing, enabeling generation of feature images on the basis of model parameters in acceptable times for a clinical application.