DOI:10.2214/AJR.04.1375
AJR 2006; 186:553-555
© American Roentgen Ray Society
Visualization and Quantification of Flow and Velocity Fields in Intracranial Arteriovenous Malformations Using Phase-Contrast MR Angiography
John W. Grinstead1,2,
Shantanu Sinha1,
Satoshi Tateshima3,4,
Yih-Lin Nien1 and
Fernando Vinuela1
1 Department of Radiological Sciences, University of California at Los Angeles,
Ueberroth Bldg., Suite 3371, 10945 LeConte Ave., Los Angeles, CA
90095-7206.
2 Siemens Medical Solutions, Los Angeles, CA 90095.
3 Department of Neurosurgery, Atsugi City Hospital, Atsugi-shi, Kanagawa, Japan,
243-0004.
4 Department of System Design Engineering, Keio University School of
Engineering, Kohoku-ku, Yohohama, Japan, 228-8522.
Received August 31, 2004;
accepted after revision April 25, 2005.
Address correspondence to J. W. Grinstead.
Abstract
OBJECTIVE. The objective of this study was to describe
postprocessing tools for MR phase-contrast flow quantification images and
apply those tools to cerebral arteriovenous malformations (AVMs) to visualize
blood flow dynamics noninvasively.
CONCLUSION. Inflow and outflow zones were clearly depicted at
different regions in the AVM. The processed images showed flow patterns
including vortical flow and variations in velocity over the cardiac cycle.
Particle tracking gave an impression of the overall flow state and of the
venous drainage system in particular.
Keywords: angiography • cine MRI • congenital malformations • hemodynamics • neuroimaging
Introduction
Most risk factors that predispose arteriovenous malformations (AVMs) to
hemorrhage affect the status of blood flow through the malformation. Mansmann
et al. [1] raised the
importance of intracranial AVM hemodynamics to predict hemorrhage risk,
although the authors mentioned that it was not easily measurable in a clinical
setting. Previous studies have already shown that MR phase-contrast flow
quantification is able to measure through-plane flow and velocity in the
cerebral arteries [2,
3]. In vivo studies of
through-plane and inplane flow dynamics gated phase-contrast flow
quantification have not been reported for intracranial AVM to our knowledge.
The objective of this study was to develop image processing tools on a modern,
flexible software platform for use with PC techniques, which will allow
noninvasive measurement of the 3D velocity distribution throughout the cardiac
cycle in cerebral AVMs.
Materials and Methods
Two volunteers were recruited for MRI and signed a consent form approved by
our institutional review board. Subject 1 was a 55-year-old man with a left
posterior frontal lobe AVM near the motor strip measuring 2 cm in greatest
diameter in the nidus and draining into superficial cortical veins. Primary
venous drainage of the AVM was a large varix that subsequently drained into
the superior sagittal sinus. Subject 2 was a 51-year-old woman with a left
occipital AVM measuring 4 cm in greatest diameter draining into both deep and
superficial venous systems.
Imaging was performed at 1.5 T (Symphony Quantum, Siemens Medical
Solutions). T2-weighted images were used to prescribe the phase-contrast flow
quantification slice in the vicinity of the nidus. The encoding velocity
(venc) was set at 100 cm/sec because maximal blood velocities
within the cerebral arteries are typically below this value even with AVMs
present. Other parameters were TE, 6.3 msec; TR between excitation pulses, 19
msec; flip angle, 30°; signal averages, 2; slice thickness, 3 mm; matrix,
256 x 240; field of view, 210 x 197 mm; triggered to 7 cine
frames; and total scanning time, 7 min 4 sec.
This study used a routine phase-contrast flow quantification pulse sequence
that has been validated in numerous studies
[2-4].
Our own validation was conducted using straight, rigid 1/4-inch-diameter tubes
connected to an MR-compatible pump with blood-mimicking fluid (CardioFlow
1000MR, Shelley Medical Imaging Technologies)
[5] and also with a
geometrically realistic in vitro aneurysm model under pulsatile flow
conditions. This model compared favorably to laser Doppler velocimetry
measurements even in the presence of nonlaminar and vortical flow
[6].
The phase-contrast flow quantification images were taken offline, sorted by
image series, and imported into Matlab (MathWorks, Inc.) for further
processing. Information in the DICOM header allowed the velocity encoding
values, encoding direction, temporal resolution, cine frame number, pixel
size, and so on to be imported automatically along with the pixel data. The
visualization tools we used and developed were previously described in more
detail [6,
7]. Briefly, considering an
axial slice, the through-plane velocity data (superior-inferior) are plotted
as contour maps with isocontour lines separating regions of differing velocity
ranges.
View larger version (154K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1A —55-year-old man with left posterior frontal lobe arteriovenous
malformation. Zoom-in shows phase image with through-plane velocity encoding.
Average was taken over all frames having through-plane encoding to improve
signal-to-noise ratio. Size of image shown is 55 x 45 mm. Points 1, 2,
and 3 show locations of measurements plotted in (B). Points 1 and 2
were placed in larger draining vein (varix) and point 3 in smaller draining
vein. In this gray-scale image, midgray is zero velocity, black is fluid
moving out of plane (superior-inferior), and white is fluid moving into plane
(inferior-superior). Particle tracking results are shown with seed points
placed at every fifth pixel.
|
|
View larger version (64K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1C —55-year-old man with left posterior frontal lobe arteriovenous
malformation. Velocity vector field superimposed onto through-plane velocity
contour map at diastole. Color bar shows velocity scale (cm/sec), with the
background color (green) being zero velocity and red and blue corresponding to
inferior-superior and superior-inferior motion, respectively.
|
|
The contour maps were colored as in Doppler sonography with blue and red
signifying flow in opposite directions. In-plane velocity data (left-right and
posterior-anterior) were superimposed as vector field maps where the arrow
length is proportional to speed. The vessels of interest were outlined
manually and only those pixels were displayed. In addition, particle paths can
be superimposed instead of vector field maps. These paths were calculated from
the phase-contrast flow quantification data using simple kinematic equations
of motion as described elsewhere
[7,
8]. Path seed points can be
chosen via mouse clicks or over an automatic range. Path calculation was
stopped if it moved outside the vessel of interest. Calculations took
approximately 80 msec per seed point on a 2.4-GHz Pentium 4 (Intel) Windows
(Microsoft) PC. Regions of interest (ROIs) at various locations throughout the
AVM were selected and the average speed was plotted over the cardiac cycle.
Importing the data and creating the figures and plots was completed in less
than 15 min.
Results
The phase-contrast image of the enlarged draining vessels of subject 1 is
shown in Figure 1A, with
particle tracking results for automatically placed seed points superimposed.
Individual seed points at specific locations can be selected by a mouse click
if desired. The software allows any combination of plots to be turned on or
off and viewed in cine mode. Flow throughout the draining veins can be seen to
follow the contours of the vessel wall. The paths connecting the varix to the
smaller draining vein are shown in error because these are separate vascular
structures having no anatomic connection. Care should be taken that each
vessel be outlined and processed separately, which can be assisted by toggling
between the anatomic and phase-contrast flow quantification images.
View larger version (87K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2 —51-year-old woman with left occipital arteriovenous malformation.
Zoom-in on sagittal slice shows dilated vein of Galen with average of
magnitude images from phase-contrast flow quantification cine acquisition
displayed. Velocity field is shown superimposed over magnitude image at
systole. Size of image shown is 35 x 35 mm.
|
|
Points 1, 2, and 3 in Figure
1A were selected by viewing the velocity cine images to find
regions with significant flow variation over the cardiac cycle where one could
visualize inflow and outflow zones. The speed at these points is plotted in
Figure 1B. The maximum speed in
the smaller draining vein at point 3 is approximately 58 cm/sec. The speeds in
the larger draining vein (points 1 and 2) are similar throughout the cardiac
cycle, and the maximum speed is a factor of 2 less than at point 3.
The contour map and velocity vector field plot are shown at diastole in
Figure 1C. The inflow and
outflow zones are clearly visualized at different regions of the AVM, with the
pair with the maximum velocities in the inferior aspect and two vortices with
two different directions of rotations in the upper part of the AVM. Velocity
changes between systole and diastole were noticeable throughout the AVM but
are most pronounced in the posterior portion and in the smaller draining vein
(point 3 in Fig. 1A). The
isocontour lines show various zones where fluid is moving into the slice
(inferior-superior, colored red) and where they are moving out
(superior-inferior, colored blue).
Figure 2 contains a sagittal
slice showing the dilated vein of Galen in subject 2. The in-plane velocity at
systole is superimposed onto the magnitude image. The vectors follow the path
of the vessel as expected, even with the slow flow that is present.
Discussion
Murayama et al. [9] could
perform intravascular sonography velocity measurements during AVM
embolotherapy because the draining vein of the animal model had a simple
anatomy, making navigation of the guidewire relatively easy. This would not
typically be feasible or justifiable in a clinical setting because draining
veins regularly have a torturous venous anatomy and a high risk during wire
navigation. Phase-contrast flow quantification can measure the flow patterns
and velocity in the feeding arteries and draining veins noninvasively. Simple
through-plane velocity measurements or 3D velocity studies of selected vessels
as shown here are possible.
The software we developed is based on a modern, extensible, and widely used
scripting platform that runs on all major operating systems. The
postprocessing is not overly time consuming but does require some experience.
The time and labor involved is similar to segmenting 3D volume images using a
semiautomated algorithm. Some modifications may be required for interpretation
in DICOM header information from different software versions because some of
the DICOM fields used in phase-contrast flow quantification are not
standardized across vendors.
In conclusion, software was developed that allows one to process
phase-contrast flow quantification images in a reasonable time on a standard
PC platform. The results show flow patterns, vortical flow features, and
variations in velocity throughout the cardiac cycle in cerebral AVMs.
References
- Mansmann U, Meisel J, Brock M, Rodesch G, Alvarez H, Lasjaunias P.
Factors associated with intracranial hemorrhage in cases of cerebral
arteriovenous malformation. Neurosurgery2000; 46:272
-279; discussion 279-281[CrossRef][Medline]
- Spilt A, Box FM, van der Geest RJ, et al. Reproducibility of total
cerebral blood flow measurements using phase contrast magnetic resonance
imaging. J Magn Reson Imaging 2002;16
: 1-5[CrossRef][Medline]
- Marks MP, Pelc NJ, Ross MR, Enzmann DR. Determination of cerebral
blood flow with a phase-contrast cine MR imaging technique: evaluation of
normal subjects and patients with arteriovenous malformations.
Radiology 1992;182
: 467-476[Abstract/Free Full Text]
- Chatzimavroudis GP, Oshinski JN, Franch RH, Walker PG, Yoganathan
AP, Pettigrew RI. Evaluation of the precision of magnetic resonance phase
velocity mapping for blood flow measurements. J Cardiovasc Magn
Reson 2001; 3:11
-19[CrossRef][Medline]
- Sinha S, Hodgson JA, Finni T, Lai AM, Grinstead J, Edgerton VR.
Muscle kinematics during isometric contraction: development of phase contrast
and spin tag techniques to study healthy and atrophied muscles. J
Magn Reson Imaging 2004; 20:1008
-1019[CrossRef][Medline]
- Tateshima S, Grinstead J, Sinha S, et al. Intraaneurysmal flow
visualization by using phase-contrast magnetic resonance imaging: feasibility
study based on a geometrically realistic in vitro aneurysm model. J
Neurosurg 2004; 100:1041
-1048[Medline]
- Grinstead J, Sinha S. In-plane velocity encoding with coherent
steady-state imaging. Magn Reson Med2005; 54:138
-154[Medline]
- Buonocore MH. Visualizing blood flow patterns using streamlines,
arrows, and particle paths. Magn Reson Med1998; 40:210
-226[Medline]
- Murayama Y, Massoud TF, Vinuela F. Hemodynamic changes in arterial
feeders and draining veins during embolotherapy of arteriovenous
malformations: an experimental study in a swine model.
Neurosurgery 1998;43
: 96-104; discussion,
104-106.[CrossRef][Medline]
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
Top
Abstract
Introduction
= 1, 
= 2, 
=
3.