Premium
Diffusion sensitivity of 3D‐GRASE in arterial spin labeling perfusion
Author(s) -
He Xiang,
Wengler Kenneth,
Schweitzer Mark E.
Publication year - 2018
Publication title -
magnetic resonance in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.696
H-Index - 225
eISSN - 1522-2594
pISSN - 0740-3194
DOI - 10.1002/mrm.27058
Subject(s) - voxel , cerebral blood flow , diffusion mri , nuclear magnetic resonance , segmentation , chemistry , biomedical engineering , biological system , computer science , artificial intelligence , physics , magnetic resonance imaging , medicine , radiology , biology , cardiology
Purpose To evaluate the role of true diffusion and flow‐related pseudodiffusion in cerebral blood flow (CBF) quantification using arterial spin labeling (ASL) with single‐shot or segmented 3D gradient and spin echo (GRASE) readouts. Theory The extended phase graph (EPG) algorithm, originally designed to model the effects of T 1 /T 2 relaxation and true diffusion in MRI acquisitions utilizing multiple refocusing RF pulses, was augmented (aEPG). This augmentation accounted for flow‐related pseudodiffusion attenuation of intravascular MRI signal in the k‐space domain during 3D‐GRASE acquisition, which leads to blur along the partition direction in the image domain. Blurring of ASL signal into neighboring voxels can lead to underestimation of CBF in small, high‐flow structures such as cortical gray matter (GM). Methods The diffusion sensitivity of 3D‐GRASE was evaluated through aEPG simulations and in vivo experiments in 13 healthy subjects. The CBF estimation bias for different postlabeling delays, crusher gradient strengths, and segmentation factors along the partition (PAR) and phase‐encoding (PE) directions was numerically assessed by simulations and experimentally validated. Results In vivo experiments demonstrated systematic underestimation of mean GM CBF measured with segmented 3D‐GRASE. The GM CBF underestimation depended on ASL preparation and imaging parameters; it reached up to 25% at low‐segmentation schemes (1 PAR × 2 PE ) but was considerably lower at high‐segmentation schemes (4 PAR × 2 PE or 8 PAR × 2 PE ). Theoretical predictions showed that conventional T 1 /T 2 relaxation and true diffusion may account for at most ∼25% of GM CBF estimation bias, whereas the pseudodiffusion effect constituted the major fraction in a typical ASL experiment. Conclusion The pseudodiffusion effect leads to considerable estimation bias in ASL‐based CBF quantification using 3D‐GRASE readouts. This bias can be substantially reduced by increasing the segmentation factors. Magn Reson Med 80:736–747, 2018. Published 2018. This article is a U.S. Government work and is in the public domain in the USA.