Florian Wiesinger^{1}, Martin A Janich^{1}, Emil Ljungberg^{1,2}, Gareth J Barker^{2}, and Ana Beatriz Solana^{1}

Here
we describe a novel method for 3D, quantitative, silent MR parameter mapping
based on 1) combined T_{1} and T_{2} magnetization preparation, 2) Zero TE image encoding
and 3) least-squares dictionary matching.

Introduction

Zero TEBecause of the low flip angles used, Zero TE results in native proton density (PD) contrast with minimal T

Here we describe a novel method for 3D, quantitative, silent MR parameter mapping based on 1) combined T

Figure
1 illustrates the pulse sequence, starting with 1) IR preparation followed by
2) six Zero TE readout segments, followed by 3) T_{2} preparation, followed by 4) another
six Zero TE readout segments. Each Zero TE
readout module, consists of N_{SpkSeg}=256 radial spokes per segment. The sequence is repeated (total number of
spokes / N_{SpkSeg}=96 times) for full 3D spatial
encoding.

The
evolution of an initial longitudinal magnetization (M_{z,0}) following
a Zero TE readout of n repetitions with flip angle α and repetition time T_{R} can be
stated as:

M_{z,n}
= M_{z,0} E_{1}^{n} cos^{n}α + M_{0}
(1- E_{1}) (1- E_{1}^{n} cos^{n}α)
/ (1- E_{1} cosα)

with
M_{0} the thermal equilibrium magnetization (i.e. proton density), and E_{1}
= e^{-TR/T1}, assuming perfect spoiling of transverse magnetization for
each T_{R}. Perfect magnetization
preparation was assumed for both inversion recovery (i.e. M_{z,n} -> -1.0*M_{z,n}) and T_{2}
preparation (i.e. M_{z,n} -> e^{-TE/T2}*M_{z,n}). The twelve Zero TE signals can be modeled as the
average of M_{z,n} over the corresponding N_{SpkSeg} spokes per
segment (cf. Figure 1).

Quantitative
PD, T_{1} and T_{2} parameter maps were obtained via pixel-wise, least-squares
matching of the measured Zero TE signals (Measure_{m}) to the
dictionary of modeled Zero TE signals (Model_{m}(T_{1},T_{2})):

Minimize: ∑_{m} | Measure_{m}(r)
– PD(T_{1},T_{2})*Model_{m}(T_{1},T_{2}) |^{2}, with
PD=∑_{m}(Measure_{m}(r)*Model_{m})/∑_{m}(Model_{m}*Model_{m}).

The
dictionary was calculated for a 384x384 equidistant grid of T_{1} (0.2s to 7s) and
T_{2} (0.01s to 1.5s) values.
The magnetization
prepared Zero TE sequence (Figure 1) was implemented on a 3T MR750w scanner (GE
Healthcare, Chicago, IL, USA) and tested in a T_{1}/T_{2} phantom^{10}
and healthy volunteers. Preparation was performed with adiabatic tanh/tan inversion and numerically
optimized T_{2} preparation designed to be robust against ±40% B_{1} variation and
±250Hz B_{0} off-resonance^{9}. Zero TE imaging parameters
were set to FOV=192mm, 1.5mm isotropic resolution, BW=±31.25kHz, FA=2°,
TR=1.4ms per spoke, 24576 total spokes per image, N_{SpkSeg}=256 spokes
per segment, scan time ~7min. Data
processing, including 3D gridding image reconstruction, least-squares
dictionary matching and visualization was done using Matlab (Mathworks, Natick,
MA).

[1] Madio et al. "Ultra‐fast imaging using low flip angles and fids." Magnetic resonance in medicine 34.4 (1995): 525-529.[2] Wu et al. "Density of organic matrix of native mineralized bone measured by water‐and fat‐suppressed proton projection MRI." Magnetic resonance in medicine 50.1 (2003): 59-68. [3] Weiger et al. "MRI with zero echo time: hard versus sweep pulse excitation." Magnetic resonance in medicine 66.2 (2011): 379-389. [4] Grodzki et al. "Ultrashort echo time imaging using pointwise encoding time reduction with radial acquisition (PETRA)." Magnetic resonance in medicine 67.2 (2012): 510-518.[5] Wiesinger, Florian, et al. "Zero TE MR bone imaging in the head." Magnetic resonance in medicine 75.1 (2016): 107-114. [6] Alibek et al. "Acoustic noise reduction in MRI using Silent Scan: an initial experience." Diagnostic and Interventional Radiology 20.4 (2014): 360.[7] Ida et al. "Quiet T1‐weighted imaging using PETRA: Initial clinical evaluation in intracranial tumor patients." Journal of Magnetic Resonance Imaging 41.2 (2015): 447-453. [8] Solana et al. "Quiet and distortion‐free, whole brain BOLD fMRI using T2‐prepared RUFIS." Magnetic resonance in medicine 75.4 (2016): 1402-1412. [9] Janich et al. “Optimal control B1-robust T2 preparation”. ISMRM 2017: 391. [10] Lerski et al. "II. Performance assessment and quality control in MRI by Eurospin test objects and protocols." Magnetic resonance imaging 11.6 (1993): 817-833. [11] Marques et al. "MP2RAGE, a self bias-field corrected sequence for improved segmentation and T 1-mapping at high field." Neuroimage 49.2 (2010): 1271-1281.

Schematic
of the magnetization prepared Zero TE pulse sequence (top), starting with
inversion recovery preparation (IRprep, red), followed by six Zero TE readout
segments each consisting of N_{SpkSeg}=256 radial spokes,
followed by T_{2} preparation (T2prep, red), followed by another six Zero TE
segments. The simulated steady-state
spin evolution (bottom), is illustrated for three different T_{1} values (T_{1}=0.5,
1, 2s, with T_{2}=0.1s, magenta) and three different T_{2} values (T_{2}=0.02, 0.1,
0.5s, with T_{1}=1s, cyan), reaching approximately SPGR steady-state (dashed
magenta lines) at the end.

Results obtained for the Eurospin phantom, showing the 12 axial images
(left), derived quantitative T_{1}, T_{2} and PD maps (middle) and comparison of
fitted vs. reference T_{1} and T_{2} values (right).

Results obtained from a healthy volunteer showing the 12 axial images
(left) and derived quantitative T_{1}, T_{2} and PD maps (middle-left). MP2RAGE
images (middle-right) and coil sensitivity maps (right) derived from the same
dataset are illustrated as well.