Donald Robinson Cantrell MD, PhD, Presenter: Nothing to Disclose

Thomas Anthony Gallagher MD, Abstract Co-Author: Nothing to Disclose

Timothy J. Carroll PhD, Abstract Co-Author: Nothing to Disclose

MR Perfusion Weighted Imaging (MR-PWI) supplements anatomic sequences by providing functional information regarding the hemodynamic status of the brain. Standard MR-PWI protocols employ singular value decomposition to deconvolve the arterial input function (AIF) from the relaxivity time curve. However, standard models are limited by their inability to account for differences in bolus arrival time (BAT), bolus dispersion, and contrast leakage. Local AIF techniques have addressed the problems associated with delayed BAT and bolus dispersion (Lee et al, 2010, MRM 63: 1305), but have not addressed leakage, an important hemodynamic parameter in the imaging of CNS malignancies (Boxerman et al, 2006, AJNR 27: 859). In the present work, we introduce a new fully automated local AIF perfusion technique that incorporates leakage effects, making it broadly applicable to both ischemia and malignancy.

A per-voxel AIF was modeled with a gamma-variate function. The residue function was modeled with a mono-exponential. Leaking contrast was defined to be proportional to the cumulative integral of the intravascular concentration time curve. Parameter values were inferred with Bayesian Markov Chain Monte Carlo simulations. The automated algorithm was implemented in C/C++ and parallelized with the Message Passing Interface to execute in parallel on a user-defined number of CPUs.

The model was applied to a pilot series of patients with Moymoya disease and CNS malignancy. In patients with Moyamoya disease, the MTT derived by standard deconvolution is confounded by bolus delay. However, the local AIF technique directly accounts for delayed BAT, and the resulting MTT prolongation is reduced compared to the standard approach. In patients with glioblastoma multiforme, tumors demonstrate variable heterogeneity with some components characterized by pronounced hyperperfusion, and other components characterized by increased vascular permeability. All of these hemodynamic properties are simultaneously inferred by our new technique.

A new local AIF technique that incorporates leakage effects has been developed, and it is broadly applicable to the evaluation of both ischemia and malignancy.

This work outlines a new local AIF technique for MR Perfusion that corrects for the effects of contrast leakage and is broadly applicable to the evaluation of both ischemia and malignancy.

Cantrell, D, Gallagher, T, Carroll, T, A New Automated Method for Magnetic Resonance Perfusion Weighted Imaging Using a Local Arterial Input Function and Contrast Agent Leakage Correction.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14010370.html