Original Research n Technical
Developments
Note: This copy is for your personal non-commercial use only. To order presentation-ready
copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.
Arterial, Venous, and
Cerebrospinal Fluid Flow:
Simultaneous Assessment with
Bayesian Multipoint Velocity-encoded
MR Imaging1
Verena Knobloch, MSc
Christian Binter, MSc
Vartan Kurtcuoglu, PhD
Sebastian Kozerke, PhD
Purpose:
To measure arterial, venous, and cerebrospinal fluid
(CSF) velocities simultaneously by using Bayesian multipoint velocity-encoded magnetic resonance (MR) imaging
and to compare interacquisition reproducibility relative to
that of standard phase-contrast MR imaging for sequential
measurements of arterial, venous, and CSF velocities.
Materials and
Methods:
This study was approved by the local ethics committee,
and informed consent was obtained from all subjects.
Simultaneous measurement of blood and CSF flow was
performed at the C1-C2 level in 10 healthy subjects (mean
age, 24.4 years 6 2.7; five men, five women) by using
accelerated Bayesian multipoint velocity-encoded MR imaging. Data were compared with those obtained from two
separate conventional phase-contrast MR imaging acquisitions, one optimized for arterial and venous blood flow
(velocity encoding range, 650 cm/sec) and the other optimized for CSF flow (velocity encoding range, 610 cm/
sec), with an imaging time of approximately 2 minutes
each. Data acquisition was repeated six times. Intraclass
correlation coefficient (ICC) and linear regression were
used to quantify interacquisition reproducibility.
Results:
There was no significant difference in arterial blood flow
measured with Bayesian multipoint velocity-encoded MR
imaging and that measured with phase-contrast MR imaging (mean ICC, 0.96 6 0.03 vs 0.97 6 0.02, respectively).
Likewise, there was no significant difference between CSF
flow measured with Bayesian multipoint velocity-encoded
MR imaging and that measured with phase-contrast MR
imaging (mean ICC, 0.97 6 0.02 vs 0.96 6 0.05, respectively). For venous blood flow, the ICC with Bayesian multipoint MR imaging was significantly larger than that with
conventional phase-contrast MR imaging (mean, 0.75 6
0.23 vs 0.65 6 0.26, respectively; P = .016).
Conclusion:
Bayesian multipoint velocity-encoded MR imaging allows
for simultaneous assessment of fast and slow flows in arterial, venous, and CSF lumina in a single acquisition. It
eliminates the need for vessel-dependent adjustment of
the velocity-encoding range, as required for conventional
sequential phase-contrast MR imaging measurements.
1
From the Institute for Biomedical Engineering, University
and ETH Zurich, Gloriastrasse 35, 8092 Zurich, Switzerland
(V. Knobloch, C.B., S.K.); the Interface Group, Institute of
Physiology, University of Zurich, Zurich, Switzerland (V.
Kurtcuoglu); and Division of Imaging Sciences and Biomedical Engineering, King’s College London, London, England.
Received April 9, 2013; revision requested May 28; revision
received July 16; accepted July 31; final version accepted
August 12. Supported by the Swiss National Science
Foundation, SmartShunt, The Hydrocephalus Project
(K-32K1_120531); and Swiss National Science Foundation
(project CR23I3_144218). Address correspondence to
S.K. (e-mail: kozerke@biomed.ee.ethz.ch).
RSNA, 2013
q
RSNA, 2013
q
566
radiology.rsna.org n Radiology: Volume 270: Number 2—February 2014
TECHNICAL DEVELOPMENTS: Simultaneous Assessment of Arterial, Venous, and Cerebrospinal Fluid Flow
Q
uantification of blood supply
and drainage of the cranium and
measurement of cerebrospinal
fluid (CSF) flow in the spinal canal
are the basis for the noninvasive assessment of intracranial compliance,
pressure (1,2), and cranial transfer
functions (3–5). These parameters
are of fundamental value in advancing the understanding and improving
treatment of cerebrospinal dynamic
disorders such as normal-pressure hydrocephalus (3) or Chiari malformations (6).
Phase-contrast magnetic resonance (MR) imaging allows for quantification of blood and CSF velocities.
The velocity-to-noise ratio of the measurement is, however, dependent on
the maximum velocity-encoding value
of the bipolar velocity-encoding gradients used in phase-contrast MR imaging. In practice, velocity encoding is
defined on the basis of estimation of
the highest velocities to be expected
in the volume of interest. Velocities
with an absolute value higher than
the velocity-encoding value will result in aliasing and require dedicated
postprocessing to unwrap the velocity
data. In contrast, velocity-encoding
values that are too high relative to
actual velocities reduce the velocityto-noise ratio proportionally and,
hence, lead to noisy measurements.
Advances in Knowledge
nn Bayesian multipoint velocityencoded MR imaging allows for
simultaneous assessment of fast
and slow flows in arterial,
venous, and cerebrospinal fluid
(CSF) lumina by using a single
protocol and provides equal or
improved interacquisition reproducibility relative to that with
conventional phase-contrast MR
imaging for arterial, venous, and
CSF flow.
nn Bayesian multipoint velocityencoded MR imaging eliminates
the need for vessel-dependent
adjustment of the velocity sensitivity, as required in conventional
phase-contrast MR imaging.
Accordingly, the issue of measurement accuracy arises if both fast- and
slow-flowing fluids are present in the
volume of interest. Although phasecontrast MR imaging has been extensively validated by using experiments
in phantoms (7–9) and in terms of
reproducibility for arterial measurements (10,11), few data are available
for domains in which large differences
in flow velocity are present.
To address the limited velocity-tonoise ratio during the slow-flow phase
of the cardiac cycle, three-point, dual
velocity-encoding, and variable velocity-encoding approaches have been
proposed (12–14). Despite these advances, assessment of intracranial
blood and CSF flow still requires separate measurements with individual
velocity-encoding values to match
the large differences in fluid velocity.
Initial studies have demonstrated simultaneous assessment by using a
dual velocity-encoding approach (15).
However, a general approach is still
lacking. In particular, venous drainage
pathways show strong intersubject
variability, with a resulting large range
of velocities (16,17). Accordingly, setting optimal velocity-encoding values
remains difficult. Even though the
assessment of venous dynamics has
been based on jugular pulsatility only
(2,9), the influence of different venous drainage pathways with different
pulsatility has been found to have a
substantial effect (18,19). Therefore,
vessels with lower velocities cannot
be neglected in venous blood flow
measurements.
To address the issue of simultaneously encoding fluid flows with large
differences in velocities, a multipoint
velocity-encoding scheme used in
combination with Bayesian processing has recently been described (20).
It has been demonstrated that such a
scheme allows for simultaneous encoding of a wide range of velocities
with an optimal velocity-to-noise ratio in a single imaging session. However, given the additional encoding
steps involved in Bayesian multipoint
measurements, the method must be
combined with data undersampling
Radiology: Volume 270: Number 2—February 2014 n radiology.rsna.org
Knobloch et al
techniques to achieve clinically feasible imaging times (21).
The objective of our study was to
measure arterial, venous, and CSF velocities simultaneously by using Bayesian multipoint velocity-encoded MR
imaging and to compare interacquisition reproducibility relative to that of
standard phase-contrast MR imaging
for sequential measurements of arterial, venous, and CSF velocities.
Materials and Methods
This prospective study was approved by
institutional and local ethics committees (Cantonal Ethics Commission Zurich) and performed during November
2012. Written informed consent was
obtained from all study participants.
Ten healthy volunteers (mean age 6
standard deviation, 24.4 years 6 2.7;
five men, five women) were enrolled in
the study. Any subjects with neurologic
disorders, including previous traumatic
brain injury, were excluded by using a
questionnaire.
Data Acquisition
Measurements were performed by using a 3.0-T system (Ingenia; Philips
Healthcare, Best, the Netherlands)
with a standard 12-element head coil
array. Data were acquired by following
the protocol shown in Figure 1. CSF
flow acquisitions in the aqueduct were
part of the protocol, but they were
not further considered in this study.
Published online before print
10.1148/radiol.13130840 Content codes:
Radiology 2014; 270:566–573
Abbreviations:
CSF = cerebrospinal fluid
ICC = intraclass correlation coefficient
Author contributions:
Guarantors of integrity of entire study, V. Knobloch, S.K.;
study concepts/study design or data acquisition or data
analysis/interpretation, all authors; manuscript drafting
or manuscript revision for important intellectual content,
all authors; manuscript final version approval, all authors;
literature research, V. Knobloch, S.K.; clinical studies,
V. Knobloch; statistical analysis, V. Knobloch, S.K.; and
manuscript editing, all authors
Conflicts of interest are listed at the end of this article.
567
TECHNICAL DEVELOPMENTS: Simultaneous Assessment of Arterial, Venous, and Cerebrospinal Fluid Flow
The preparation for section planning
included survey imaging and time-offlight angiography. A transverse imaging section was prescribed at the
C1-C2 level for the quantification of
blood and CSF flow. Bayesian multipoint velocity-encoded MR imaging data with six velocity-encoding
points and one reference point were
acquired with fivefold undersampling
in k-space and time with principal
component analysis reconstruction.
Taking into account the acquisition of
15 training profiles for k-t principal
component analysis reconstruction,
a net acceleration factor of 3.5 was
achieved. Accordingly, the total imaging time for a Bayesian multipoint encoding measurement was identical to
that for single conventional two-point
phase-contrast MR imaging. As reference, two conventional phase-contrast
MR imaging data sets were collected
in separate measurements before and
after Bayesian multipoint acquisition
with velocity encodings of 50 and 10
cm/sec, respectively. To test measurement reproducibility, subjects were
asked to stand up for a few minutes
after the first set of three measurements. The protocol was repeated
once after repositioning. Heart rates
before and after repositioning did not
change substantially. By using prospective electrocardiographic triggering, measurement times were dependent on heart rate; ranges for imaging
time are given in Figure 1.
Imaging parameters were as follows: gradient-echo imaging; field of
view, 180 3 145 mm; matrix size, 224
3 180; section thickness, 5 mm; repetition time, 11 msec; and echo time,
7.2 msec. Velocity encodings were interleaved over cardiac cycles, and three
k-space profiles were acquired consecutively, resulting in a temporal resolution of three times the repetition time
(ie, 33 msec). Depending on heart rate,
20–29 cardiac phases were recorded by
using prospective electrocardiographic
triggering.
Data Processing
Data processing was implemented by
using software (Matlab; Mathworks,
568
Knobloch et al
Figure 1
Figure 1: Measurement protocol. After three repetitions of arterial (A), venous (V), and CSF
velocity acquisition interleaved with two repetitions of aqueductal CSF flow, volunteers were asked
to stand up and walk around. Protocol was restarted after repositioning. Imaging times for phasecontrast MR imaging (PC-MRI) and multipoint acquisitions were dependent on heart rate. RARE =
rapid acquisition with relaxation enhancement, TOF = time of flight, venc = velocity encoding.
Figure 2
Figure 2: Data processing of data acquired at C1-C2 level. Left: Phase-contrast MR imaging (PC-MRI) and
multipoint data were reconstructed individually and segmented in the same way. k-t = k-t space and time,
kv = velocity k-space, PCA = principal component analysis. Right: Segmentation template threshold was set
during first imaging, and vessels were numbered according to vessel scheme shown. Arteries are in red and
veins are in blue. Threshold level was adapted at subsequent imaging to ensure identical vessel sizes.
Natick, Mass). Data were analyzed by a
reader (V. Knobloch) with 3 years of experience in MR flow imaging. An overview
of processing steps is given in Figure 2.
The phase-contrast MR imaging data
were reconstructed by using standard
radiology.rsna.org n Radiology: Volume 270: Number 2—February 2014
TECHNICAL DEVELOPMENTS: Simultaneous Assessment of Arterial, Venous, and Cerebrospinal Fluid Flow
Knobloch et al
Figure 3
Figure 3: MR images
allow visual comparison of velocities (v) at
different velocity scaling.
Top: Images are scaled
to 670 cm/sec to show
arterial and jugular vein
velocities. Insets are
enlargements of internal
carotid artery and jugular
vein. Bottom: Images are
scaled to 610 cm/sec to
highlight CSF velocities
in spinal canal. Insets are
enlargements of spinal
canal. PC-MRI = phasecontrast MR imaging, venc
= velocity encoding.
Fourier transformation and phase difference processing. Phase aliasing was
corrected by using the Matlab unwrap
algorithm applied along the temporal
dimension. Multipoint data were reconstructed by using k-t principal component analysis (21), and velocities were
reconstructed by using the Bayesian reconstruction algorithm (20). Each data
set was segmented individually to account for volunteer motion between the
two acquisitions. Segmentation of multipoint and phase-contrast MR imaging
data was based on the modified angiographic image, A, as follows:
A = σ [ I (t )] ⋅ ∑ v ⋅ ∑ mag,
T
T
where s[I(t)] is the standard deviation
over time of the complex image I, v the
velocity, mag the image magnitude, and
T the number of cardiac phases per cycle. In the first multipoint acquisition
of each set (Fig 1), arteries and veins
were segmented by manually setting
the threshold, and the number scheme
shown in Figure 2 was manually assigned as the original vessel mask. The
segmentation was automatically adapted to all multipoint and phase-contrast
MR images (velocity encoding = 50 cm/
sec) for each vessel, as follows: The angiographic image was masked with a dilated vessel mask of the original image.
A threshold was computed to result in
the same vessel size as that on the original mask. The spinal canal was manually segmented for each acquisition.
Velocity profiles and stroke volumes
were computed for each vessel and
each acquisition. Stroke volume of the
spinal canal was computed as the sum
of forward and backward stroke volumes. As a measure of reproducibility,
intraclass correlation coefficient (ICC)
3,1 (22) was computed for each vessel.
The three repeat acquisitions with
Bayesian multipoint MR imaging and
Radiology: Volume 270: Number 2—February 2014 n radiology.rsna.org
with standard phase-contrast MR imaging were correlated by using multiple linear regression. Velocities for
each cardiac phase and in each vessel
were correlated among and across the
two sets of three measurements each.
Deviation of the linear regression line
from the identity line between measurements is given in degrees and by
the intercept (ie, the distance of the
fitted line from zero). The standard
error of the estimate was computed as
the mean distance of the data relative
to the line obtained by using linear regression. Normalization relative to the
mean vessel velocity was performed.
Results
Velocity images from phase-contrast
MR imaging and multipoint data at the
time point of maximal CSF flow in the
spinal canal acquired at 188 msec after
the R wave are shown in Figure 3. The
569
TECHNICAL DEVELOPMENTS: Simultaneous Assessment of Arterial, Venous, and Cerebrospinal Fluid Flow
Figure 4
Figure 4: Graphs of ICC as a function of peak velocity (vpeak) for arteries, veins, and CSF (ie, spinal canal).
Means 6 standard deviations for all values are shown in bar graphs on right. PC-MRI = phase-contrast MR
imaging, venc = velocity encoding.
top row is scaled to 70 cm/sec to visualize arterial velocities, and the lower
row is scaled to 10 cm/sec to visualize CSF velocities. Velocities obtained
with the multipoint sequence showed
good velocity-to-noise ratios for both
arterial and CSF flow. In contrast, CSF
velocities acquired with phase-contrast
MR imaging exhibited amplified noise
when velocity encoding was 50 cm/
sec. Arterial and venous velocities
were wrapped multiple times at phasecontrast MR imaging with velocity
encoding of 10 cm/sec, as expected,
and required additional unwrapping in
postprocessing.
Figure 4 summarizes reproducibility
given as the ICCs for all vessels of all
subjects as a function of peak velocity.
570
In arteries, no significant difference was
found between Bayesian multipoint MR
imaging and phase-contrast MR imaging with a velocity encoding of 50 cm/
sec (mean ICC, 0.96 6 0.03 vs 0.97
6 0.02, respectively), whereas significantly higher ICCs were seen for multipoint imaging in veins (mean ICC, 0.75
6 0.23 vs 0.65 6 0.26; P = .016). Flow
of CSF in the spinal canal was detected
equally well with multipoint MR imaging and phase-contrast MR imaging
with a velocity encoding of 10 cm/sec
(mean ICC, 0.97 6 0.02 vs 0.96 6 0.05,
respectively). In general, it was noted
that multipoint acquisitions resulted in
improved performance, particularly for
low peak velocities as found in venous
and CSF compartments.
Knobloch et al
Results of linear regression analysis
of velocities determined in the three
repeat acquisitions are reported as
the difference in slope, intercept, and
standard error of the estimate. Arterial velocities were found to be close to
identity for both multipoint and phasecontrast MR imaging with a velocity encoding of 50 cm/sec (mean difference in
slope: 3.2° 6 2.3 vs 2.7° 6 2.0, respectively; mean intercept: 1.8 cm/sec 6
0.9 vs 1.7 cm/sec 6 1.4, respectively).
The mean standard error of the estimate was 5% 6 2 for both multipoint
and phase-contrast MR imaging. For
venous velocities, the mean difference
in slope was 13.2° 6 13.1 for multipoint MR imaging and 12.9° 6 13.1 for
phase-contrast MR imaging with a velocity encoding of 50 cm/sec; the mean
intercept was 4.2 cm/sec 6 5.9 and 3.0
cm/sec 6 3.6, respectively. There was
a significant reduction in the standard
error of the estimate for multipoint MR
imaging compared with phase-contrast
MR imaging (mean, 8.1% 6 4.5 vs 20%
6 15.2, respectively; P , .001). Linear
regression of CSF velocities resulted in
a mean differnce in slope of 2.6° 6 1.7
for multipoint MR imaging and 7.3° 6
5.8 and 3.7° 6 4.6 for phase-contrast
MR imaging with velocity encodings of
50 and 10 cm/sec, respectively. The
mean intercept of CSF was 0.1 cm/
sec 6 0.1 with multipoint MR imaging and 0.3 cm/sec 6 0.2 and 0.1 cm/
sec 6 0.1 for phase-contrast MR imaging with a velocity encodings of 50
and 10 cm/sec, respectively. Significant
differences were found with regard to
the standard error of the estimate between multipoint and phase-contrast
MR imaging with a velocity encoding of 10 cm/sec (mean, 17.0% 6 6.0
vs 11.8% 6 4.2, respectively; P ,
.001) and between multipoint and
phase-contrast MR imaging with a velocity encoding of 50 cm/sec (mean,
17.0% 6 6.0 vs 37.5% 6 14.7, respectively; P , .001).
Stroke volumes for all arteries and
veins are shown in Figure 5. There
were no significant differences between
phase-contrast MR imaging and multipoint acquisition. The intrasubject
standard deviation was considerably
radiology.rsna.org n Radiology: Volume 270: Number 2—February 2014
TECHNICAL DEVELOPMENTS: Simultaneous Assessment of Arterial, Venous, and Cerebrospinal Fluid Flow
smaller than the intersubject standard
deviation in all vessels (P , .001). Both
inter- and intrasubject standard deviations were higher for veins than for
arteries (P , .001).
The correlation matrix between all
vessels and CSF in the spinal canal is
shown in Figure 6. Correlation between
arteries was stronger than intervenous
correlation (P , .001). CSF velocities
correlated mostly positively with those
of the venous system.
Knobloch et al
Figure 5
Discussion
In this study, we demonstrated the simultaneous assessment of arterial, venous, and CSF velocities with use of
Bayesian multipoint velocity encoding
and showed it could provide values similar to those obtained with conventional
phase-contrast MR imaging. Overall,
Bayesian multipoint velocity encoding
was found to provide similar or improved interacquisition reproducibility
relative to that of sequential phase-contrast MR imaging, with the added advantage of requiring only half the total
imaging time. The technique facilitates
flow measurements in clinically feasible
times.
Although conventional phase-contrast
MR imaging requires phase unwrapping if
the magnitude of actual velocities exceeds
velocity encoding, data from multipoint
encoding is inherently free of aliasing as
long as actual velocities are lower than
or equal to the maximum of the entire
encoding range, which was 104 cm/sec
in our study. Accordingly, a single protocol is sufficient to map a large dynamic
range, simplifying imaging setup and minimizing planning overhead.
The ICC as a measure of reproducibility was found to be high for arterial
velocities, with a slight reduction at
lower peak velocities. For venous velocities, single values of ICC were low at
higher peak velocities. This is related to
reduced pulsatility of venous flow and
can also be associated with physiologic
factors. It has been shown that venous
flow is dependent on posture (23) and
respiration (24). Because posture was
supine during the entire examination,
there was no direct influence expected
Figure 5: Bar charts show stroke volumes (SV) for all arteries, veins, and CSF (ie, spinal canal). Numbers
in parentheses are numbers of subjects in whom the respective vessel could be identified. PC-MRI = phasecontrast MR imaging, s = standard deviation.
Figure 6
Figure 6: Correlation matrix of velocities in all arteries, veins, and CSF.
from this end. However, changes in respiratory patterns may have occurred,
possibly influencing venous flow. Respiratory monitoring was not performed, and, thus, data on respiratory
Radiology: Volume 270: Number 2—February 2014 n radiology.rsna.org
variations during imaging were not recorded in our study.
Stroke volumes were similar for
vessels on the left versus right side of
the neck. Stroke volumes in the left
571
TECHNICAL DEVELOPMENTS: Simultaneous Assessment of Arterial, Venous, and Cerebrospinal Fluid Flow
and right internal carotid arteries
(1,2) were higher than those in vertebral arteries (3,4). Venous drainage was mainly through the left and
right jugular veins, but other pathways were also identified. Correlation
matrix analysis revealed differences
in pulsatility. The highest correlation was found between intracranial
arteries. In the venous system, the
following regions of higher correlation
were identified: Jugular veins exhibited high correlation between the left
and right sides, whereas vertebral
veins and spinal column venous drainage correlated positively with that of
arteries. Another cluster of strong
correlation was identified for veins
positioned posteriorly to the spinal
canal. CSF flow was found to correlate positively with that of the jugular veins while showing negative correlation with arterial velocities. This
is in line with previous findings (18).
However, to confirm the venous correlation dependencies, studies in larger
cohorts are warranted.
Interindividual variations in venous pulsatility patterns are associated with certain disorders (25). Differences in anatomy and pulsation can
be linked to increased outflow resistances, as found in chronic cerebrospinal venous insufficiency (26). The
simultaneous assessment of all venous
outflow pathways, together with arterial and CSF flow, is also considered
important for the noninvasive assessment of intracranial pressure (2). Because pulsatility can vary in amplitude
and phase, the acquisition of flow
waveforms in all veins is important for
a complete description of intracranial
pulsation.
This work was a proof-of-concept
and feasibility study of measuring
blood and CSF velocities simultaneously by using Bayesian multipoint
velocity-encoded MR imaging. This
initial study was limited in that only
healthy young adults were investigated. Accordingly, further studies are
warranted to translate the approach
to relevant patient populations.
In summary, Bayesian multipoint
velocity-encoded MR imaging allows
572
for simultaneous assessment of arterial, venous, and CSF velocities with
equal or improved interacquisition
reproducibility relative to that of sequential phase-contrast MR imaging
acquisitions and holds potential for
studies in patients.
Disclosures of Conflicts of Interest: V. Knobloch No relevant conflicts of interest to disclose. C.B. No relevant conflicts of interest to
disclose. V. Kurtcuoglu No relevant conflicts
of interest to disclose. S.K. No relevant conflicts of interest to disclose.
References
1.Alperin N, Mazda M, Lichtor T, Lee SH.
From cerebrospinal fluid pulsation to noninvasive intracranial compliance and pressure measured by MRI flow studies. Curr
Med Imaging Rev 2006;2(1):117–129.
2. Alperin NJ, Lee SH, Loth F, Raksin PB, Lichtor T. MR-intracranial pressure (ICP): a
method to measure intracranial elastance
and pressure noninvasively by means of
MR imaging—baboon and human study.
Radiology 2000;217(3):877–885.
3.Balédent O, Gondry-Jouet C, Meyer ME,
et al. Relationship between cerebrospinal
fluid and blood dynamics in healthy volunteers and patients with communicating hydrocephalus. Invest Radiol 2004;39(1):45–
55.
4.Schmid Daners M, Knobloch V, Soellinger M, et al. Age-specific characteristics and coupling of cerebral arterial inflow and cerebrospinal fluid
dynamics.
PLoS
ONE
2012;7(5):
e37502.
5.Tain RW, Alperin N. Noninvasive intracranial compliance from MRI-based measurements of transcranial blood and CSF
flows: indirect versus direct approach.
IEEE Trans Biomed Eng 2009;56(3):544–
551.
6. Alperin N, Sivaramakrishnan A, Lichtor T.
Magnetic resonance imaging–based measurements of cerebrospinal fluid and blood
flow as indicators of intracranial compliance in patients with Chiari malformation.
J Neurosurg 2005;103(1):46–52.
7. Greil G, Geva T, Maier SE, Powell AJ. Effect
of acquisition parameters on the accuracy
of velocity encoded cine magnetic resonance imaging blood flow measurements.
J Magn Reson Imaging 2002;15(1):47–54.
8. Lotz J, Meier C, Leppert A, Galanski M. Cardiovascular flow measurement with phasecontrast MR imaging: basic facts and im-
Knobloch et al
plementation. RadioGraphics 2002;22(3):
651–671.
9. Wåhlin A, Ambarki K, Hauksson J, Birgander R, Malm J, Eklund A. Phase contrast
MRI quantification of pulsatile volumes of
brain arteries, veins, and cerebrospinal
fluids compartments: repeatability and
physiological interactions. J Magn Reson
Imaging 2012;35(5):1055–1062.
10.Spilt A, Van den Boom R, Kamper AM,
Blauw GJ, Bollen EL, van Buchem MA.
MR assessment of cerebral vascular response: a comparison of two methods. J
Magn Reson Imaging 2002;16(5):610–616.
11. Koerte I, Haberl C, Schmidt M, et al. Inter- and intra-rater reliability of blood and
cerebrospinal fluid flow quantification by
phase-contrast MRI. J Magn Reson Imaging 2013;38(3):655–662.
12.Ringgaard S, Oyre SA, Pedersen EM.
Arterial MR imaging phase-contrast flow
measurement: improvements with varying
velocity sensitivity during cardiac cycle.
Radiology 2004;232(1):289–294.
13.Buonocore MH. Blood flow measurement
using variable velocity encoding in the RR
interval. Magn Reson Med 1993;29(6):
790–795.
14.Lee AT, Pike GB, Pelc NJ. Three-point
phase-contrast velocity measurements
with increased velocity-to-noise ratio.
Magn Reson Med 1995;33(1):122–126.
15.Dhoondia HA, Alperin N. Improved MRintracranial pressure (MR-ICP) measurement using a new data acquisition
technique [abstr]. In: Proceedings of the
eleventh meeting of the International Society for Magnetic Resonance in Medicine.
Berkeley, Calif: International Society for
Magnetic Resonance in Medicine, 2003;
793.
16.Doepp F, Schreiber SJ, von Münster
T, Rademacher J, Klingebiel R, Valdueza JM. How does the blood leave
the brain? A systematic ultrasound
analysis of cerebral venous drainage patterns. Neuroradiology 2004;
46(7):565–570.
17.Stoquart-Elsankari S, Lehmann P, Villette
A, et al. A phase-contrast MRI study of
physiologic cerebral venous flow. J Cereb
Blood Flow Metab 2009;29(6):1208–1215.
18. Marshall I, MacCormick I, Sellar R, Whittle I. Assessment of factors affecting
MRI measurement of intracranial volume
changes and elastance index. Br J Neurosurg 2008;22(3):389–397.
19.Pomschar A, Koerte I, Lee S, et al. MRI
evidence for altered venous drainage and
radiology.rsna.org n Radiology: Volume 270: Number 2—February 2014
TECHNICAL DEVELOPMENTS: Simultaneous Assessment of Arterial, Venous, and Cerebrospinal Fluid Flow
intracranial compliance in mild traumatic
brain injury. PLoS ONE 2013;8(2):e55447.
20. Binter C, Knobloch V, Manka R, Sigfridsson
A, Kozerke S. Bayesian multipoint velocity
encoding for concurrent flow and turbulence
mapping. Magn Reson Med 2013;69(5):
1337–1345.
21. Pedersen H, Kozerke S, Ringgaard S, Nehrke K, Kim WY. k-t PCA: temporally constrained k-t BLAST reconstruction using
principal component analysis. Magn Reson
Med 2009;62(3):706–716.
22.Shrout PE, Fleiss JL. Intraclass correla
tions: uses in assessing rater reliability.
Psychol Bull 1979;86(2):420–428.
23.Alperin N, Lee SH, Sivaramakrishnan A,
Hushek SG. Quantifying the effect of posture on intracranial physiology in humans
by MRI flow studies. J Magn Reson Imaging 2005;22(5):591–596.
24.Zamboni P, Menegatti E, Pomidori L,
et al. Does thoracic pump influence the
cerebral venous return? J Appl Physiol
2012;112(5):904–910.
Radiology: Volume 270: Number 2—February 2014 n radiology.rsna.org
Knobloch et al
25.Beggs CB. Venous hemodynamics in neurological disorders: an analytical review
with hydrodynamic analysis. BMC Med
2013;11:142.
26. Zamboni P, Galeotti R, Menegatti E, et al.
Chronic cerebrospinal venous insufficiency
in patients with multiple sclerosis. J Neurol Neurosurg Psychiatry 2009;80(4):392–
399.
573