Roberto Colotti1, Jean Delacoste1, Giulia Ginami1, Maxime Pellegrin2, Tobias Kober1,3,4, Yutaka Natsuaki5, David Grodzki6, Ulrich Flögel7, Matthias Stuber1,8, and Ruud B. van Heeswijk1
1Department of Radiology, University Hospital (CHUV) and University of Lausanne (UNIL), Lausanne, Switzerland, 2Division of Angiology, University Hospital of Lausanne (CHUV), Lausanne, Switzerland, 3Advanced Clinical Imaging Technology, Siemens Healthcare IM BM PI, Lausanne, Switzerland, 4LTS5, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland, 5Siemens Medical Solutions, NAM USA DI MR COLLAB WE, Los Angeles, CA, United States, 6Siemens Healthcare GmbH, HC DI MR R&D PLH, Erlangen, Germany, 7Department of Cardiovascular Physiology, Heinrich Heine University, Düsseldorf, Germany, 8Center for Biomedical Imaging (CIBM), Lausanne, Switzerland
Synopsis
Fluorine-19 (19F) MRI of perfluorocarbon emulsions (PFCs) with
multiresonant spectra is challenging due to destructive phase interference that
leads to short T2 relaxation times (<10 ms). Pulse sequences with
very short echo times (≤ 100 µs) can be used to overcome this challenge. In
this study, in vitro and in vivo 19F MRI obtained with both UTE and PETRA were acquired and quantitatively compared.Introduction
Fluorine-19 (
19F) MRI of perfluorocarbon
emulsions (PFCs) is increasingly used for inflammation studies due to its high
specificity and non-toxicity [1]. The PFC perfluorooctyl bromide (PFOB) has already
been used in clinical trials as a blood volume expander [2] and oxygen carrier
[3]. However, PFOB has a complex multiresonant spectrum (Fig.1), in which the
different resonances J-couple to one another, which causes their apparent T
2
relaxation time to shorten significantly. This
imaging hurdle can be overcome by acquiring the signal before destructive spin
interference occurs, for example with ultra-short echo (UTE) and pointwise encoding time
reduction with radial acquisition (PETRA) pulse sequences.
In this study, we compared the signal
to noise ratio (SNR) and the image quality of a UTE
pulse sequence that consists of a 3D radial trajectory with balanced
steady-state free precession (bSSFP) acquisition [4], with PETRA, which
consists of a gradient echo (GRE) 3D radial half-projection outer k-space combined
with a pointwise-acquired Cartesian inner k-space.
Methods
For the phantom study, a 50mL tube was filled with 2%
agar and with five 1mL syringes with 2% agar gel and PFOB [5] at different
concentrations (0-5.9 M). All experiments were performed on a 3T clinical system
(Magnetom Prisma, Siemens Healthcare, Germany). A 35-mm-diameter volume transmit/receive coil tuneable on both
19F/
1H
resonances (Rapid Biomedical, Rimpar,
Germany) was used for excitation
and signal detection. For
19F MRI, the excitation frequency was set
to the CF
2(δ-ε) resonance, while the 3D radial bSSFP-UTE pulse sequence was used
with the following parameters: field of view (FOV)=160×160×160mm3, repetition
time TR=1.94ms, echo time
TE=50µs,
pixel bandwidth BW=1202Hz/pixel, RF excitation angle 30°, isotropic voxel
volume=1mm
3, radial views=125952, acquisition time=4min10s. A prototype 3D GRE PETRA pulse sequence was tested
with the following parameters: TR=3.29ms, TE=40µs, BW=1838Hz/pixel, RF
excitation angle 6°, effective voxel volume=0.1 mm
3, reconstructed
voxel volume=1 mm
3, radial views=126000, acquisition time=7min14s.
1H images were acquired with
the same respective pulse sequences and resolution as the
19F images.
The
SNR of the
19F images of the different PFOB tubes was then
determined and normalized to the voxel volume and acquisition time. Here, the
SNR of a region of interest (ROI) was defined as the ratio between the average
signal in the ROI and the standard deviation in a background ROI of at least 20x20
pixels outside the object of interest. In both cases, a linear fit of the
19F concentration versus the SNR
was performed, and the quality of the fit was compared. For a quantitative
assessment of image quality, the edge blurring (EB) [6] of the tube with
highest concentration was calculated. For in vivo validation, three 13-week-old apolipoprotein-E-knockout
(ApoE-/-) C57BL/6 mice
(well-established model for vascular inflammation) were intravenously injected with 300µL of PFOB. Seven
days after the injection, MRI was performed. After
1H
anatomical reference scans,
19F UTE and PETRA images centered on the abdomen
were acquired with the same parameters as the phantom scans.
1H and
19F
images were acquired with the same respective pulse sequences and resolution. All
animal studies were approved by the local animal ethics committee. The normalized
SNR from
19F images was compared between the two pulse sequences.
Results
For
19F
images, the normalized SNR versus concentration linear fit for the UTE pulse
sequence was characterized by a 1.7 times higher
slope (and thus sensitivity) than that from the PETRA pulse sequence (Fig.3). The
EB was 2.50±0.12 pixels for PETRA and 2.91±0.16 pixels for UTE pulse sequence
(p<0.001). In the phantoms, UTE showed more ringing artifacts than PETRA for
both
1H and
19F images (Fig.2). The goodness of the linear fit was R
2=0.95
for both UTE and PETRA.
19F signal was detected in vivo in the spleen
of all animals with both pulse sequences (Fig.4), and the normalized SNR was slightly
higher for PETRA when compared to the UTE pulse sequence (31.2±4.9 vs. 29.6±5.3).
Discussion and Conclusions
Both the UTE and PETRA pulse sequences have ultra-short
echo times that helped suppress the destructive dephasing arising from the
coupled PFOB resonances. In vitro, the sensitivity was 1.7 times higher for UTE,
while PETRA allowed for a higher pixel bandwidth and this resulted in higher
image quality as evidenced by the lower edge blurring. However, in vivo, PETRA demonstrated
slightly higher normalized SNR, which might be due to better robustness to motion
of the animal and to different physiological environment of PFOB.
Acknowledgements
Swiss National Science Foundation Grant PZ00P3_154719
(RBvH),
Fondation Pierre Mercier (RBvH)
References
[1] Ruiz-Cabello et al., NMR Biomed 24(2):114-29, 2011
[2] Riess et al., Artif Cells
Blood Substit
Immobil Biotechnol 34(6):567-580, 2006
[3] Riess et al., Chem Rev 101(9):2792-2919,
2001
[4] Goette et al., Magn Reson Med 74(2):537-543,
2014
[5] Jacoby et al., NMR Biomed 27(3):261-271, 2014
[6]
Kording et al., Magn Reson Med
74(5):1257-1265, 2014.