Effects of flip angle uncertainty and noise on the accuracy of DCE-MRI metrics: comparison between standard concentration-based and signal difference methods.

Document Type

Article

Abstract

Dynamic contrast-enhanced MRI is becoming an increasingly important tool to assess tumors and their response to treatment. In the most common method of computing tumor perfusion parameters, the concentration of the injected contrast agent is first computed in both tumor and blood which is subsequently fit to a perfusion model, typically the Tofts two compartment model. However, this strategy can be highly sensitive to errors in the excitation flip angle and noise. More recently, a simpler method of determining perfusion was developed in which the difference signal, obtained by subtracting the measured time course signal by the signal prior to bolus arrival, is utilized in lieu of the concentration values. The goal of this work is to compare the performance of these two strategies with simulation experiments in the presence of flip angle errors and different levels of image signal to noise ratios (SNRs). Results show that with the conventional method, if assumed pre-contrast T1 of blood is used, large errors in perfusion (exceeding 400% and 200% for K(trans) and ve, respectively) can occur in the presence of flip angle deviations typically observed in vivo. However, when baseline T1 values are measured for both tumor and blood, the errors become a function of flip angle difference between the two locations, with nearly no error if the flip angle errors are identical at both locations. The errors are substantially smaller with the signal difference strategy (less than 100% for both K(trans) and ve). The latter method also yields more consistent perfusion values at varying SNR levels. The results suggest that measuring the actual flip angle may be critical for obtaining absolute perfusion values, but in studies in which relative changes in perfusion is of primary interest or if true flip angles are not known, the signal difference strategy may be preferred over the standard concentration-based method.

Keywords

Algorithms, Computer Simulation, Contrast Media, Humans, Image Processing, Computer-Assisted, Imaging, Three-Dimensional, Magnetic Resonance Imaging, Models, Statistical, Perfusion, Reproducibility of Results, Signal Processing, Computer-Assisted, Signal-To-Noise Ratio, Uncertainty

Medical Subject Headings

Algorithms; Computer Simulation; Contrast Media; Humans; Image Processing, Computer-Assisted; Imaging, Three-Dimensional; Magnetic Resonance Imaging; Models, Statistical; Perfusion; Reproducibility of Results; Signal Processing, Computer-Assisted; Signal-To-Noise Ratio; Uncertainty

Publication Date

1-1-2015

Publication Title

Magnetic resonance imaging

ISSN

1873-5894

Volume

33

Issue

1

First Page

166

Last Page

173

PubMed ID

25311569

Digital Object Identifier (DOI)

10.1016/j.mri.2014.10.005

This document is currently not available here.

Share

COinS