Patrick Winter1, Kristina Andelovic2, Thomas Kampf1,3, Peter Jakob1,4, Wolfgang Bauer2, and Volker Herold1
1Experimentelle Physik V, Universität Würzburg, Würzburg, Germany, 2Medizinische Klinik und Poliklinik I des Universitätsklinikums Würzburg, Würzburg, Germany, 3Universitätsklinikum Würzburg, Institut für Diagnostische und Interventionelle Radiologie, Würzburg, Germany, 4Fraunhofer IIS, Fraunhofer EZRT, Magnetresonanz- und Röntgenbildgebung (MRB), Universität Würzburg, Würzburg, Germany
Synopsis
A
self-navigated radial 4D-PC sequence is presented for accelerated ECG-free 4D
flow measurements in the murine aortic arch. Self-navigation signals were
extracted from the radial DC signal and used for retrospective motion
synchronization. 3D-Cines with 30 frames were reconstructed with a spatial
resolution of 100 µm. The volume flow was determined at 4 2D slices extracted
from the 3D dataset and the 3D flow was visualized with streamlines. The
results are in good accordance with results reported for ECG-triggered
measurements. The new method yields high potential for preclinical studies of
hemodynamics and can also be transferred to applications in humans.
Purpose
Flow quantification using 4D PC-Cine MRI became recently an important
tool for preclinical studies of atherosclerosis, which enables non-invasive measurements
of cardiovascular hemodynamics as well as the assessment of functional
parameters such as the aortic compliance and endothelial shear stress. Major
disadvantages of previously introduced triggered 4D flow quantification
techniques are the very long measurement time needed to achieve the necessary
high temporal and spatial resolutions and the dependency on stable cardiac
motion synchronization signals. Especially in small rodents flow measurements
are challenging due to the difficult animal handling and the ECG signal’s
vulnerability to errors at high magnetic field strengths. In this abstract we
present a new self-navigated 4D PC-Cine MRI technique that enables
high-resolution (100 µm isotropic) 4D flow measurements of the complete murine
aortic arch in 32 minutes. The use of retrospective reconstruction and
self-gating enables a very flexible data analysis and increases the robustness
since an ECG signal is no longer needed for cardiac motion synchronization.Materials and Methods
MR measurements
All measurements are carried out on a 17.6T small
animal MRI scanner with a 1 T/m gradient system and a 24 mm birdcage coil. Mice
were anesthetized with isoflurane (1.5-2% in oxygen) and kept on a constant body
temperature of 37°C. For the 3-dimensional flow encoding a radial 4D-PC-MRI
sequence with a balanced 4-point encoding sequence1 was used. The
acquisition parameters were: VENC=125 cm/s, TR/TE=3.0/1.1 ms, echo asymmetry:
10%, 160000 spokes, FOV: 25x25x4 mm3. For slice excitation a
sinc-shaped pulse with a flip angle of 15° was used. The 4D flow measurements
were conducted without triggering during free breathing and required a total
scan time of 32 minutes.
Reconstruction
Cardiac and respiratory motion signals were
extracted from the radial DC signal. High frequency noise was suppressed with a
matched filter2 and a baseline
correction was applied. The processed self-gating signals were afterwards used
for breath gating and for the assessment of trigger time stamps and the
relative positions in the heart cycle, as described recently3,4. The
information about the cardiac phases was used for the subsequent sliding window
(window width: 1/30 of the cardiac period) selection of the radial projections
needed for the 4D-Cine reconstruction.
For each velocity encoding step 3D images with
an isotropic spatial resolution of 100x100x100 µm3 were
reconstructed with convolution gridding5 at 30 different cardiac phases. All
reconstructions were done with Matlab (The Mathworks, Inc., Natick, USA) and
the velocity information was visualized with Ensight (CEI Software, USA).
Results
Self-navigated 4D flow
measurements were conducted in the murine aortic arch. Fig. 1 shows an example
of a self-gating signal measured in an ApoE-/- mouse. The
modulations due to cardiac motion (red arrows) and respiration (blue arrows)
are clearly recognizable. The mean cardiac period during this measurement was
(106.4±3.4) ms. Fig. 2 illustrates a 3D isosurface of the arch at the systolic
cardiac phase together with maximum intensity projections of the magnitude
image. Four slices were extracted from the 3D dataset to determine the volume
flow through the aorta. The results can be seen in Fig. 3 (from slice 1:
ascending aorta to slice 4: descending aorta). The peak flow decreases for
slices with larger distances to the aortic root due to the partial blood flow
into the major branches. The flow values are in good accordance with previous
studies6. Fig. 4 shows a streamline illustration of the blood flow through
the aortic arch and the smaller arteries.
Discussion and Conclusion
In this abstract we demonstrate
the feasibility of accelerated self-navigated 4D flow measurements in the mouse
with very high temporal (30 frames / cardiac cycle) and spatial resolutions
(100 µm). Since flow measurements in the aortic arch are achievable in 32
minutes the coverage of the complete aorta is possible in about 1 hour. The
proposed method does not require ECG signals for motion synchronization and
should hence lead to higher robustness and an increased animal handling. The
retrospective approach allows a very flexible data analysis that can be
optimized to increase either temporal or spatial resolution. Possible
applications of this new method are studies of the hemodynamics in
atherosclerotic mouse models, 3D measurements of the aortic pulse-wave-velocity
and of the endothelial shear stress. This
method can also be transferred to applications in humans.
Acknowledgements
This work was supported by
grants from the Deutsche Forschungsgemeinschaft (SFB 688 B5, Z2), the Bundesministerium
für Bildung und Forschung (BMBF01 EO1004) and the Comprehensive Heart Failure Center (CHFC).References
[1] Pelc et al., Jour
Magn Reson Med, [1991]; 405-413
[2] Herold et al., Proc. ISMRM 2016; 0466
[3] Winter et
al., JCMR, [2013], 15:88-98
[4] Winter et al., Magn Reson Med, [2016]; DOI: 10.1002/mrm.26068
[5] Fessler JA; IEEE Trans Signal Process
2003;51(2):560–574.
[6] Janizcek et
al., Magn Reson Med, [2011]; 66:1382-1390