Meeting Banner
Abstract #0554

Estimation of Pharmacokinetic Parameters in Dynamic Contrast Enhanced MRI via Random Forest Regression

Cagdas Ulas1, Michael J. Thrippleton2, Ian Marshall3, Mike Davies4, Paul A. Armitage5, Stephen D. Makin2, Joanna M. Wardlaw2, and Bjoern H. Menze1

1Computer Science, Technical University of Munich, Munich, Germany, 2Neuroimaging Sciences, University of Edinburgh, Edinburgh, United Kingdom, 3Centre for Clinical Brain Sciences, University of Edinburgh, Edinburgh, United Kingdom, 4Institute for Digital Communication, University of Edinburgh, Edinburgh, United Kingdom, 5Cardiovascular Science, University of Sheffield, Sheffield, United Kingdom

We propose a novel alternative approach to estimate pharmacokinetic (PK) parameters of dynamic contrast enhanced (DCE)-MRI. Our approach leverages machine learning field and mainly targets to automatically learn temporal patterns of the voxel-wise concentration-time curves (CTCs) from a large amount of training samples in order to make accurate parameter estimations. We consider the estimation of parameters as a regression problem and specifically use Random Forest (RF) regression. We demonstrate its potential and utility to improve the conventional model-fitting based quantitative analysis of DCE-MRI especially in various noise conditions, and validate our method on clinical brain stroke datasets.

This abstract and the presentation materials are available to members only; a login is required.

Join Here