Dahan Kim1,2, Carson Hoffman1, Oliver Wieben1,3, and Kevin M. Johnson1
1Department of Medical Physics, University of Wisconsin, Madison, WI, United States, 2Department of Physics, University of Wisconsin, Madison, WI, United States, 3Department of Radiology, University of Wisconsin, Madison, WI, United States
Synopsis
In this work, we examined the feasibility of registering 4D-flow MRI scans with different scan sequences, and demonstrate how the incorporation of complementary, registered data can enhance characterization of hemodynamic information. Black blood (BB) and 4D-flow magnitude images demonstrated excellent registration between the pre- and post-rotation data sets, with high values of correlation and good overlap of the vessels between head rotation. Joint visualization of aneurysm 4D flow and BB shows accurate lesion depiction only after registration and is a promising technique for the comprehensive evaluation of vascular pathology.Purpose
4D-Flow MRI enables the visualization of
blood flow and quantification of hemodynamic parameters for characterizations
of cardiovascular diseases (CVD). However, the analysis and interpretation of
4D-Flow data alone is often challenging.
For example, vessel segmentation is required for quantification and the
frequently used PC angiogram can become unreliable in cases of slow or irregular
flow patterns (1). Contrast-enhanced MR angiogram offers improves
vessel segmentation but is not co-registered to the 4D-flow data. Further, new interpretation of flow features
may be enhanced by the incorporation of alternative contrasts, such as arterial
wall enhancement (2, 3) superimposed on 4D flow data. The
purpose of this work is to develop an imaging visualization paradigm that
harnesses the synergistic information of a multi-contrast vascular exam. Paramount to this technique is collection of
high-resolution 3D vascular scans and accurate registration before the joint
visualization, which may be challenging in cases of dissimilar contrast.
Methods
To evaluate the registration accuracy of
dissimilar MRI images, a set of T1 weighted black blood (BB) and 4D-Flow intracranial
scans were performed twice on four healthy volunteers as shown in Figure 1. First, BB and 4D-Flow scans were acquired
back to back with the subject instructed to remain as still as possible (BB
1
and Flow
1). Subsequently,
subjects were asked to rotate their head (~5-10°) and the scans were repeated
(BB
2 and Flow
2). Imaging
was performed on a 3T scanner (MR750, GE Healthcare, WI, USA) with a 32channel
head coil (Nova Medical, MA, USA). Sagittal BB images were collected utilizing
a DANTE (4) prepared spin echo sequence with
0.75x0.8x0.8mm
3 resolution.
4D-Flow was collected with an axial 3D radially undersampled sequence(5) with 0.7mm isotropic
spatial resolution. 3D rigid image
registration was performed with open-source software (Advanced Normalization
Tools) using a mutual information metric. First, BB
1 and BB
2
were registered to their counterpart Flow
1 and Flow
2.
Next, images before and after the head rotation were registered using
dissimilar images (e.g. BB
1-Flow
2 and Flow
1-BB
2).
Within the inherent registration error
due to interpolation and noise, the registration between Flow
1 and
Flow
2 should be equivalent to the registration involving two
dissimilar registrations (e.g. Flow
1-BB
1-Flow
2).
Quality of the registration between
rotations was assessed via correlation of PC angiograms (Flow
1 and
Flow
2), as the
correlation is not a suitable measure for registration quality between
dissimilar image series such as BB and 4D-Flow.
Registration was incorporated into a
multi-contrast vascular workup, including 4D-Flow, CE-MRA, and black blood
images in a patient scan. After
preprocessing, images were visualized jointly in a commercial flow
visualization package (EnSight, CEI, Apex, USA) before and after registration
of a subject with an
intracranial aneurysm. Due to the
extended exam time (~40min), interscan motion was expected. All images were
registered to the 4D-Flow scan.
Results
BB
and 4D-Flow magnitude images demonstrated excellent registration between the
pre- and post-rotation data sets, as shown in Figure 2. The correlation between
the pre- and post-rotation angiograms (averaged over all four volunteers) increased
from 0.167±0.081 before the registration to
0.810±0.117 after the registration, similar to correlation using the
direct registration of the two 4D-Flow images: 0.874±0.058. (Imperfect
ρ<1 is partially due to non-rigid motion of superficial temporal arteries).
When the dissimilar registration was
applied to the aneurysm patient, the overlay of the bright vessels from the 4D
flow scan matches with the black “lumens” of the dark vessels in the BB images
(Fig. 3, right), instead of tissues (Fig. 3, left). Joint visualization of the
brain aneurysm with the 4D flow MRI and BB sequence shows that blood flow is
accurately displayed within the vessels of BB only after the dissimilar
registration between 4D flow MRI and BB sequence.
Discussion
Satisfactory registration between the
pre- and post-rotation data sets, as visualized in Fig. 2 and quantified by
increased correlations above, indicates that dissimilar registration using BB
and 4D-Flow images can identify and match anatomical similarities between the
two seemingly different images (Fig. 1). As shown with the aneurysm patient
(Fig. 3 and 4), joint visualization of 4D-Flow MRI and other diagnostic images
in everyday clinical settings must be preceded by dissimilar registrations
between them, as done in this study. This
analysis platform may be useful in studies investigating the complex
relationships between hemodynamic conditions and vascular remodeling in cases
of aneurysms, atherosclerosis, and arterial venous malformations.
Acknowledgements
We gratefully acknowledge funding from NIH-NS066982
and GE Healthcare and for research support.References
1. Huston
J, Rufenacht DA, Ehman RL, Wiebers DO. Imtracranial Aneurysms and Vascular
Malformations: Comparison of Time-of-Flight and Phase-Contrast Angiography.
Radiology. 1991;181(3):721-30.
2. Edjlali M, Gentric JC,
Regent-Rodriguez C, Trystram D, Ben Hassen W, Lion S, Nataf F, Raymond J,
Wieben O, Turski P, Meder JF, Oppenheim C, Naggara O. Does Aneurysmal Wall
Enhancement on Vessel Wall MRI Help to Distinguish Stable From Unstable
Intracranial Aneurysms? Stroke. 2014;45(12):3704-6. doi:
10.1161/Strokeaha.114.006626. PubMed PMID: WOS:000345516600266.
3. Hasan DM, Hesitad D. Early Change in
Ferumoxytol-Enhanced Magnetic Resonance Imaging Signal Suggests Unstable Human
Cerebral Aneurysm. A Pilot Study. Stroke. 2013;44(2). PubMed PMID:
WOS:000330540200106.
4. Li L, Miller KL, Jezzard P.
DANTE-prepared pulse trains: a novel approach to motion-sensitized and
motion-suppressed quantitative magnetic resonance imaging. Magn Reson Med.
2012;68(5):1423-38. doi: 10.1002/mrm.24142. PubMed PMID: 22246917.
5. Johnson
KM, Lum DP, Turski PA, Block WF, Mistretta CA, Wieben O. Improved 3D phase
contrast MRI with off-resonance corrected dual echo VIPR. Magn Reson Med.
2008;60(6):1329-36. PubMed PMID: 19025882.