Marijn van Stralen1, Esben Thade Petersen1,2, Jeroen Hendrikse1, and Clemens Bos1
1University Medical Center Utrecht, Utrecht, Netherlands, 2Danish Research Center for MR, Hvidovre, Denmark
Synopsis
Abdominal perfusion
imaging using contrast media injection is potentially nephrotoxic. Arterial
spin labeling (ASL), employing endogenous contrast, was shown using spatially
selective labeling strategies. We investigated the reproducibility of velocity
selective ASL (VS-ASL), which eliminates delicate label planning and possibly
improves perfusion SNR by labeling closer to the target tissue. We show that abdominal
VS-ASL is feasible in healthy volunteers and overcome labeling artifacts by
pacing and triggering the acquisition with good temporal SNR. However, VS-ASL
is sensitive to motion during readout, deteriorating reproducibility. It could
benefit from outlier rejection techniques and retrospective motion correction. Purpose
Abdominal
perfusion imaging of the kidneys using CT or MRI contrast media is costly and potentially
nephrotoxic. These limitations have sparked the desire to develop abdominal
perfusion techniques that do not require contrast media injection. Arterial
spin labeling (ASL) MRI is a candidate technique that employs the endogenous
contrast of labeled protons in the blood to image abdominal perfusion.
Previously, the feasibility of abdominal ASL has been investigated using spatially
selective, pulsed (PASL) and pseudo-continuous (pCASL) labeling strategies on
the kidneys
1,2 and liver
3. Recently, feasibility velocity selective ASL
(VS-ASL)
4 was investigated for the kidneys
5 to eliminate the delicate aspects
of planning the spatially selective label slab. VS-ASL also potentially
improves SNR by labeling closer to the tissue of interest, which reduces the
time of the label to reach the target tissue and thus preserving its signal.
However, this recent study
5 called for motion compensation strategies to
suppress labeling artifacts due to typical abdominal motion. Therefore, we
investigated the reproducibility of pCASL and VS-ASL for renal and hepatic
perfusion mapping in healthy volunteers under different triggering conditions.
Methods
Imaging
Five healthy volunteers (mean
age 32yo, range [22-47], 2 female) were scanned on a 1.5T MR system (Ingenia,
Philips Healthcare, Best, the Netherlands), equipped with a 32-channel torso
coil. Written informed consent was obtained. Volunteer preparation included
intake of 500ml juice to suppress susceptibility artifacts near the stomach and
duodenum. In each exam, three coronal ASL scans were acquired twice; 1. pCASL
scan with labeling slab positioned on the aorta and a fixed TR of 6500ms, 2.
VS-ASL with a BIR-4 based tagging pulse6 and a fixed TR of 6500ms (fVS-ASL),
and 3. tVS-ASL, triggered on expiration (tVS-ASL).
Volunteers were instructed to shortly prolong their end-expiratory state to
match the fixed TR of 6500ms, allowing for in- and expiration after the
acoustically easily recognizable readout, such that labeling and readout were
performed in exhaled position.
The velocity selective labeling employed a
velocity threshold of 2.5 cm/s in the craniocaudal direction and a post labeling
delay of 500ms. Pseudo-continuous labeling duration was 1450ms.
Coronal
imaging was performed with a multi-slice single shot gradient-echo EPI Look-Locker
readout, with four repetitions for pCASL (post-label delay 50ms) and five for
VS-ASL (post-label delay 500ms).
Imaging parameters were TE 4.7ms, acq matrix
96x86, 3.9x4.0mm, 14 slices, slice thickness 8mm, gap 1.0mm, parallel imaging
factor 2.5, EPI factor 57, flip angle 30°, for all ASL scans, with 20
repetitions of label and control pairs.
To confirm constant perfusion
conditions, phase-contrast flow measurements on the left renal artery were
interleaved with the ASL scans throughout the scan session. Anatomical T1 and
T2 weighted scans were acquired for segmentation of the liver and kidneys.
Analysis
The kidneys and liver were
manually segmented. To assess the consistency of the labeling for each of the
methods, temporal SNR (temporally averaged signal intensity/temporal standard
deviation) was measured for the 20 control-label repetitions. Pairwise (control-label)-subtractions
were averaged and perfusion maps were calculated by fitting the data to the
general kinetic model for Look-locker readout7 and expressed as ml/100ml/min.
Reproducibility of the VS-ASL techniques was assessed for the liver and
kidneys. All processing was implemented using MeVisLab (v2.6.2, MeVis Medical
Solutions AG, Bremen, Germany).
Results
Temporal
SNR (tSNR) measurements (Fig. 1) did not reveal significant differences between
the labeling methods on any of the regions of interests. Overall tSNR (mean
± SD) were 33.5
± 2.0, 31.6
± 2.1 and 32.5
± 2.7 for pCASL, fVS-ASL and tVS-ASL
respectively. Renal artery flow measurements confirmed constant perfusion
conditions over the scan sessions. Although the feasibility of abdominal VS-ASL
was confirmed by good visual correspondence (Fig. 2 and 3), VS-ASL
reproducibility measurements were hindered by motion artifacts caused by
differences in exhalation level at readout.
Discussion
This
study on the use of abdominal VS-ASL in healthy volunteers confirms the
feasibility of the technique. Although motion during labeling was avoided by
pacing the volunteers breathing cycle or triggering the labeling, differences
in exhalation level at readout were still present and hindered reproducibility
measurements. However, these breathing motion artifacts might be suppressed by retrospective
motion correction. This is under investigation. Future research will also focus
on rejection strategies and comparison of the resulting perfusion maps with
DCE-MRI and vascular input flow measurements.
Conclusions
Abdominal
perfusion imaging using VS-ASL is feasible and promising. In order to improve
the reproducibility of the technique, future research should be directed
towards retrospective motion correction to suppress motion artifacts introduced
during readout.
Acknowledgements
We
would like to thank MeVis Medical Solution AG (Bremen, Germany) for the use of
MeVisLab in this study for image analysis.References
1 Martirosian et al. MRM 2004;51(2);353–361
2 Robson PM et al. MRM 2009;61(6);1374-1387
3 Schalkx et al. Eur Radiol 2015;25(6);1529-1540
4 Wong et al. MRM 2006;55(6);1334-1341
5 Van
Stralen et al. ISMRM 2015
6 Wong EC & Guo J ISMRM 2010
7 Guenther et al. MRM 2001;46(5);974-984