Improving perfusion measurement in DSC-MR imaging with multiecho information for arterial input function determination

Document Type

Conference Proceeding

Abstract

BACKGROUND AND PURPOSE: Clinical measurements of cerebral perfusion have been increasingly performed with multiecho dynamic susceptibility contrast-MR imaging techniques due to their ability to remove confounding T1 effects of contrast agent extravasation from perfusion quantification. However, to this point, the extra information provided by multiecho techniques has not been used to improve the process of estimating the arterial input function, which is critical to accurate perfusion quantification. The purpose of this study is to investigate methods by which multiecho DSC-MRI data can be used to automatically avoid voxels whose signal decreases to the level of noise when calculating the arterial input function. MATERIALS AND METHODS: Here we compare postprocessing strategies for clinical multiecho DSC-MR imaging data to test whether arterial input function measures could be improved by automatically identifying and removing voxels exhibiting signal attenuation (truncation) artifacts. RESULTS: In a clinical pediatric population, we found that the Pearson correlation coefficient between R2 time-series calculated from each TE individually was a valuable criterion for automated estimation of the arterial input function, resulting in higher peak arterial input function values while maintaining smooth and reliable arterial input function shapes. CONCLUSIONS: This work is the first to demonstrate that multiecho information may be useful in clinically important automatic arterial input function estimation because it can be used to improve automatic selection of voxels from which the arterial input function should be measured.

Keywords

AIF=arterial input function, QM=quality of merit, R2=change in effective transverse relaxation rate, R=Pearson correlation coefficient, RMSerror=root-mean-square error

Publication Date

7-1-2016

Publication Title

American Journal of Neuroradiology

ISSN

01956108

E-ISSN

1936959X

Volume

37

Issue

7

First Page

1237

Last Page

1243

PubMed ID

26988812

Digital Object Identifier (DOI)

10.3174/ajnr.A4700

Share

COinS