Assessment of MRI contrast agent concentration by quantitative susceptibility mapping (QSM): application to estimation of cerebral blood volume during steady state

Abstract

ObjectiveOne major issue in dynamic susceptibility contrast MRI (DSC-MRI) is to accurately determine contrast agent (CA) concentration, since T2* relaxivity in vivo is generally unknown and varies between blood and tissue. In this study, quantitative susceptibility mapping (QSM) was used for quantification of CA concentration.Materials and methodsA DSC-MRI protocol, including phase data acquisition, was applied to 20 healthy volunteers in a test–retest study. By selecting a CSF reference region of interest (ROI), the values of all QSM images were shifted to show no CA-induced change in CSF. CA concentration and cerebral blood volume (CBV) were estimated using shifted QSM data. CSF reference ROI optimization was evaluated by investigation of CBV repeatability. The CBV age dependence was analysed and tissue T2* relaxivity was estimated.ResultsThe best repeatability of CBV, using an optimal CSF reference ROI, showed test-versus-retest correlations of r = 0.81 and r = 0.91 for white and grey matter, respectively. A slight CBV decrease with age was observed, and the estimated in vivo T2* relaxivity was 85 mM−1s−1.ConclusionProvided that a carefully selected CSF reference ROI is used to shift QSM image values, susceptibility information can be used to estimate concentration of contrast agent and to calculate CBV.Electronic supplementary materialThe online version of this article (doi:10.1007/s10334-017-0637-9) contains supplementary material, which is available to authorized users.

Measurements of cerebral blood volume using quantitative susceptibility mapping, R2* relaxometry, and ferumoxytol‐enhanced MRI

“…Another study implemented QSM‐ and R 2 *‐based approaches to estimate CBV. The authors report values of 2.93 ± 1.33% in WM and 4.22 ± 1.52% GM from QSM‐based estimates, and 7.4 ± 3.6% in WM and 14.7 ± 7% GM from R 2 *‐based estimates . Both the QSM‐ and the R 2 *‐based CBV results they report are larger than those in this study.…”

“…The authors report values of 2.93 ± 1.33% in WM and 4.22 ± 1.52% GM from QSM-based estimates, and 7.4 ± 3.6% in WM and 14.7 ± 7% GM from R 2 *-based estimates. 23 Both the QSM-and the R 2 *-based CBV results they report are larger than those in this study. That study was performed using a GBCA, a single-echo (T E = 29 ms) DSC-MRI sequence and a 60°flip angle.…”

“…The conjugate gradient tolerance was 0.01 and maximum number of iterations 100. Based on the phantom experiments, and other in vivo studies, the regularization parameter λ = 300 was selected for the in vivo data analysis . QSM maps were generated using an AMD Opteron 6134 CPU with 28 CPU cores and 120 GB memory, running in a virtual machine with CentOS Linux 7.…”

“…Based on the phantom experiments, and other in vivo studies, the regularization parameter λ = 300 was selected for the in vivo data analysis. 23,26 QSM maps were generated using an AMD Opteron 6134 CPU with 28 CPU cores and 120 GB memory, running in a virtual machine with CentOS Linux 7. To account for the sensitivity of QSM measurements to an arbitrary constant shift, zero referencing was performed using CSF measurements from QSM images.…”

“…However, dynamic QSM was also found to underestimate CBF, which was attributed to imperfect background field removal and deconvolution of GBCA concentration throughout time with the arterial input function. Another DSC‐MRI study measured CBV from contrast agent concentrations using QSM and found age dependent trends . However, the authors noted an underestimation of gray matter (GM) to white matter (WM) CBV ratios.…”