Am J Physiol Heart Circ Physiol 315: H1174 –H1181, 2018.
First published July 20, 2018; doi:10.1152/ajpheart.00017.2018.
RESEARCH ARTICLE Translational Physiology
Normal patterns of thoracic aortic wall shear stress measured using four-
dimensional flow MRI in a large population
Fraser M. Callaghan1,2 and Stuart M. Grieve1,2,3,4,5
1
Sydney Translational Imaging Laboratory, Heart Research Institute, Charles Perkins Centre, University of Sydney, Sydney,
New South Wales, Australia; 2Sydney Medical School, University of Sydney, Camperdown, New South Wales, Australia;
3
Department of Radiology, Royal Prince Alfred Hospital, Camperdown, New South Wales, Australia; 4Macquarie Medical
Imaging, Macquarie University Hospital, Macquarie, New South Wales, Australia; and 5Sydney Adventist Hospital,
Wahroonga, New South Wales, Australia
Submitted 9 January 2018; accepted in final form 18 July 2018
Callaghan FM, Grieve SM. Normal patterns of thoracic aortic
wall shear stress measured using four-dimensional flow MRI in a
large population. Am J Physiol Heart Circ Physiol 315: H1174 –
H1181, 2018. First published July 20, 2018; doi:10.1152/ajpheart.00017.
2018.—Wall shear stress (WSS) plays a governing role in vascular
remodeling and a pathogenic role in vessel wall diseases. However,
little is known of the normal WSS patterns in the aorta as there is
currently no practical means to routinely measure WSS and no normal
ranges derived from population data exist. WSS measurements were
made on the aorta of 224 subjects with normal anatomy using
four-dimensional flow MRI with multiple encoding velocities and an
optimized postprocessing routine. The spatial and temporal variation
in WSS and oscillatory shear index was analyzed using a flat map
representation of the unfolded aorta. The influence of aortic shape and
velocity on WSS was evaluated using regression analysis. WSS in the
thoracic aorta is dominated by axial flow. Average peak systolic WSS
was 1.79 ⫾ 0.71 Pa in the aortic arch and was significantly higher at
2.23 ⫾ 1.04 Pa in the descending aorta, with a strong negative
correlation with advancing age. The spatial distribution of WSS is
highly heterogeneous, with a localized region of elevated WSS along
the length of the anterior wall seen across all individuals. Our data
demonstrate that accurate four-dimensional flow-derived WSS mea-
surement is feasible, and we further provide a standardized parametric
approach for presentation and analysis. We present a normal range for
WSS across the lifespan, demonstrating a decrease in WSS with
advancing age as well as illustrating the high degree of spatial and
temporal variation.
NEW & NOTEWORTHY With the use of four-dimensional flow
MRI and postprocessing, accurate direct measurement of wall shear
stress (WSS) was performed in a population of normal thoracic aortas
(n ⫽ 224). WSS was higher in the descending aorta compared with the
aortic arch and decreased with age. A heterogeneous pattern of
elevated WSS along the length of the aorta anterior wall was consis-
tent across the population. This work provides normal data across the
adult age range, permitting comparison with pathology.
aorta; fluid dynamics; four-dimensional flow; magnetic resonance
imaging; wall shear stress
Address for reprint requests and other correspondence: S. M. Grieve,
Sydney Translational Imaging Laboratory, Heart Research Institute, Sydney
Medical School and Charles Perkins Centre, Univ. of Sydney, Camperdown
2006, NSW, Australia (e-mail: stuart.grieve@sydney.edu.au).
H1174 0363-6135/18 Copyright © 2018 the American Physiological Society
Downloaded from journals.physiology.org/journal/ajpheart at Kangwon Natl Univ (203.252.078.033) on May 22, 2022.
INTRODUCTION
Abnormal flow dynamics play a role in the initiation and
progression of aortic diseases such as atherosclerosis, aneu-
rysms, and dissections, but only limited means exist to directly
measure their effects (4, 21, 22). One of the important patho-
logical consequences of abnormal flow patterns is the genera-
tion of wall shear stress (WSS) magnitudes or spatial and
temporal WSS patterns that fall outside of an “optimal range”
that is suited to the anatomy, wall structural, and cellular
characteristics of the aorta. As such, accurate measurements of
WSS values represent a major gap in current cardiovascular
phenotyping capability. Here, we present the first clinically
meaningful description of WSS across a population with nor-
mal aortic dimensions and cardiac function.
Evidence for the links among aberrant aortic flow, abnormal
WSS, and aortic pathological processes stems from observa-
tions of the common geographic patterns of aortic disease. For
example, atherosclerotic plaques typically form in characteris-
tic locations, often occurring in regions of flow recirculation at
the inner wall of bifurcations or around the ostium of the arch
vessels in the aorta (10, 11). Similarly, aortic dissections are
more common in regions of known high-flow disturbance, such
as the ascending aorta or immediately after the left subclavian
vessel (27). These striking geographical phenomena are backed
up by animal and in vitro data in controlled experimental
systems, which demonstrate that macroscopic cellular and
molecular wall alterations are causally related to exposure to
WSS outside of well-defined limits (19). Despite the strong
consensus formed by clinical observations and these direct
basic measurements, little confirmatory direct data exist in
patient populations to support the proposal that abnormal WSS
is causally associated with disease progression in patients, and
there is an absence of any population data to inform normal
ranges of WSS (23).
In the absence of a readily applicable means to directly
measure WSS, computational fluid dynamics (CFD) has been
applied to help understand both the spatial and temporal
patterns of WSS and also the influence of aortic flow dynamics
on this parameter (3). Four-dimensional flow (4D-flow) MRI is
a technique able to measure the dynamic velocity field in the
heart and major blood vessels. It has been used extensively to
observe, quantify, and calculate hemodynamic parameters such
as high velocity jets, flow rates, vorticity, helicity, and energy
http://www.ajpheart.org
 DIRECT FOUR-DIMENSIONAL FLOW MEASUREMENT OF WSS IN THE THORACIC AORTA
transfer in the aorta (2, 9, 16). Previous WSS magnitudes in the
normal aorta measured using 4D-flow MRI have been reported
to be in the order of 0.5–1.0 Pa, although these numbers are
acknowledged to be underestimates in light of the spatial
limitations of estimating WSS using standard 4D-flow MRI
data (20, 30). In contrast, WSS magnitudes in the normal aorta
reported from CFD studies are consistently reported in the
range 2.0 –5.0 Pa (3, 18). To overcome this limitation of
4D-flow, we adopt a previously validated postprocessing tech-
nique capable of increasing near-wall resolution and thus
improving WSS measurement accuracy (6).
WSS is the product of friction between moving fluid and the
vessel wall and is proportional to the change in fluid velocity
moving away from the wall surface. The natural curvature of
the aorta introduces secondary flow dynamics and asymmetry
to the bulk flow motion (14, 15). The flow dynamics and shape
of the aorta therefore directly influence WSS distribution and
magnitude. With advancing age, the aorta enlarges, the arch
changes shape from a near-perfect semicircle, and the vessel
generally becomes more tortuous (28, 32). Understanding the
influence of age-related changes in aortic shape and flow
dynamics on WSS is therefore important.
In this report, we present the results of highly optimized
4D-flow-derived WSS measurements in a large cohort, repre-
senting the spectrum of the different shapes and aortic size
present in the population without aortic pathology. Our data
provide the most comprehensive depiction of the normal range
of WSS in the native aorta to date and will enable the use of
experimentally derived ranges for comparison purposes.
METHODS
Study cohort. This was a prospective multicenter investigation
(4D-biomarker) designed to apply novel analytical techniques to
understand basic physiological processes using 4D-flow. Participants
provided written informed consent in accordance with the ethical
guidelines of the relevant institutional review boards (Macquarie
Medical Imaging, Macquarie University Hospital, Macquarie, NSW,
Australia; Department of Radiology, Sydney Adventist Hospital,
Wahroonga, NSW, Australia; and Department of Radiology, Royal
AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00017.2018 • www.ajpheart.org
Downloaded from journals.physiology.org/journal/ajpheart at Kangwon Natl Univ (203.252.078.033) on May 22, 2022.
H1175
Prince Alfred Hospital, Camperdown, NSW, Australia). Subjects
included those undergoing MRI scans as part of routine clinical
assessment for suspected or confirmed cardiovascular disease as well
as healthy control subjects recruited to the study. Inclusion criteria
were as follows: a full 4D-flow multivelocity encoding (VENC) data
set and the presence of informed consent. Exclusion criteria were as
follows: age ⬍ 18 yr old (n ⫽ 15), prior aortic surgery or known aortic
pathology (n ⫽ 18), abnormal aortic valve (regurgitant fraction ⬎ 3%,
aortic stenosis or a bicuspid valve), and moderate or severe left
ventricular dysfunction (ejection fraction ⬍ 40%).
MRI acquisition. Data were acquired at three separate sites on two
types of MRI scanners: a GE Medical Systems 3T 750W MRI or a
Siemens 3T Skyra (Erlangen, Germany). Acquisition used an 18-
channel phased array anterior coil in combination with a linear
12-channel posterior coil. For each subject, two 4D-flow data sets
were acquired using a four-point referenced phase-encoding strategy
at subject-specific encoding velocities of 150 –170 cm/s (high VENC)
and at 50 – 60 cm/s (low VENC), determined from two-dimensional
phase-contrast measurements made at the ascending aorta (24). 4D-
flow data were acquired using 2.5-mm sagittal slices covering the
entire heart and great vessels. Other parameters were as follows: GE
acquisitions: echo time of 1.98 and 2.18 ms, repetition time of 5.34
and 8.72 ms (high and low VENCs respectively), and resolution of
2.5-mm isotropic; Siemens acquisitions: echo time of 2.8 and 3.6 ms,
repetition time of 5.14 and 5.8 ms (high and low VENCs respec-
tively), and resolution of 2.5-mm isotropic. The acquired temporal
resolution was 47 ⫾ 12 ms (21 ⫾ 2.5 phases), dependent on subject
heart rate and scanner limitations, across both prospective (Siemens)
and retrospectively (GE) ECG-gated free-breathing sequences. Accel-
eration factors used were GRAPPA factor ⫽ 3 (Siemens) and k-t
acceleration factor ⫽ 8 (GE). All subjects had standard cardiac MRI
“cine” (2- and 4-chamber and short axis) views and two-dimensional
phase-contrast sequences acquired to assess for cardiovascular dis-
ease. Total MRI scan time was ~45–55 min.
4D-flow postprocessing. Data were processed offline using in-
house code written in python and C⫹⫹ and using the open source
library VTK (29). Processing was performed on a Linux Rocks
computer cluster comprising 28 CPUs over 5 nodes (Intel i7, 32 GB
DRR4 RAM). Data were corrected for background phase effects and
localized phase aliasing. Multiple VENC acquisitions were combined
giving a higher velocity to noise ratio at low velocities using a
previously described technique (8). Finally, the VFIT denoising algo-
Fig. 1. Semiautomatic isolation of aortic lu-
men from the level of the main pulmonary
artery bifurcation to the level of the dia-
phragm from the four-dimensional flow MRI
velocity magnitude data [coronal (A) and
sagittal views (B)]. C and D: definition of
shape factors [radius of curvature of the
aortic arch (AoCurve; C) and aortic shape
(AoShape; D)].
H1176 DIRECT FOUR-DIMENSIONAL FLOW MEASUREMENT OF WSS IN THE THORACIC AORTA

Fig. 2. A representative subject showing the three-dimen-

sional aorta lumen, colored by peak systolic wall shear
stress (WSS) and the corresponding “unwrapped” flat map
of the lumen cut along a line the length of the aorta defining
the most posterior point at each axial division, relative to
the plane of aortic arch curvature, denoted as 12 o’clock
in the flat map view. Arrows about the circumference of the
aorta demonstrate the direction of unwrapping beginning
from the cut line, with markings of 3 o’clock and 6 o’clock
labeled. **Corresponding location in the flat map.

rithm was applied, which serves to improve the 4D-flow data spatial point. This allowed a parametric representation of WSS and its
resolution, eliminate noisy peaks, and improve the accuracy of near- derivatives across all subjects for simple analysis and interindividual
wall velocity measurements to improve WSS calculations (6). Post- comparison (Fig. 2). At each parametric point, the magnitude of the
processing produced a cleaned time-resolved vector field representing WSS vector, axial, and circumferential components, with respect to
flow throughout the aorta over the cardiac cycle at an interpolated the local polar coordinate system of the aorta cross section, were
resolution of 1.1–1.5 mm (6). defined.
Aortic parameterization and curvature calculation. For each sub- In addition, time-averaged WSS (tWSS) and oscillatory shear index
ject, the aorta was characterized by geometric and fluid dynamic (OSI) were calculated. OSI represents the transient orientation of
parameters likely to influence WSS measures. An aortic centerline WSS over a cardiac cycle defined by the ratio of tWSS to the average
was defined manually, beginning at the left ventricular outflow track WSS vector (AWSSV), as follows:
and extending to the descending aorta beyond the level of the
1
兰
T
diaphragm. The aortic centerline was then used to extract the aortic tWSS ⫽ 㛳WSS㛳dt (1)
lumen via a custom, automatic routine based on local inspection of the T 0

冉 冊
flow vectors in a plane perpendicular to the centerline. Following this
1 AWSSV
process, each aortic lumen surface was inspected manually to ensure OSI ⫽ 1⫺ (2)
accurate segmentation. The final aortic surface was used to recalculate 2 tWSS
a refined accurate aortic centerline (1). Due to movement of the aortic
root and ascending aorta, WSS measurements were limited to the where
portion of the thoracic aorta beginning at the level of the pulmonary 1
兰
T
bifurcation, extending to the diaphragm (Fig. 1, A and B). AWSSV ⫽ 㛳 WSSdt㛳 (3)
T 0
Aortic arch curvature was defined by extracting the aortic arch
portion of the centerline (6 cm proximal and distal from the most where T is the time period of a cardiac cycle.
superior point of the aortic centerline) and generating a plane of best Statistical analysis. Continuous variables are expressed as
fit of these centerline points and then performing a least-squares circle means ⫾ SD. Comparisons were made using an unpaired t-test, with
fitting to the aortic arch centerline points projected onto this plane, significance expressed for two-tailed P ⬍ 0.05. Linear regression
thus providing a radius of curvature (AoCurve; Fig. 1C). analyses were applied with details specific to each analysis described
The angle between the centerline point at the crest of the transverse
aortic arch and centerline points 6 cm proximal and distal to this was
calculated to define the aortic arch shape (AoShape; Fig. 1D). An
AoShape angle of 180° thus represents a purely square/crenel shape; Table 1. Clinical, demographic, and cardiac function
90° is a perfectly round curve with a diameter of 6 cm, and angles description
below this represent increasingly triangular or “gothic” shapes (with
an extreme value of 0 being an impossible, vertical arch). Parameter Means ⫾ SD
WSS analysis on thoracic aorta. WSS was calculated at the aortic n 224
lumen at each 4D-flow acquired time step using a B-spline fit to the Age, yr 49.58 ⫾ 17.97
local, VFIT-corrected velocity profile and assuming a constant blood Men/women, % 68/32
viscosity of 3.5 mPa·s. The aortic lumen surface was parameterized Height, m 1.36 ⫾ 0.73
into 100 evenly spaced axial divisions with the proximal 40 divisions Weight, kg 79.54 ⫾ 20.60
termed the aortic arch and the distal 60 divisions termed the descend- Body surface area, m2 1.53 ⫾ 0.84
ing aorta. At each axial division, a local polar coordinate system was Left ventricular ejection fraction 62.68 ⫾ 12.09
Left ventricular end-diastolic volume index 74.44 ⫾ 22.01
established perpendicular the aortic centerline and divided into 24
Left ventricular stroke volume, ml 89.33 ⫾ 26.98
divisions, clockwise with respect to the primary (proximal to distal)
flow direction such that “12 o’clock” aligned with the most posterior Values are means ⫾ SE.

AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00017.2018 • www.ajpheart.org

Downloaded from journals.physiology.org/journal/ajpheart at Kangwon Natl Univ (203.252.078.033) on May 22, 2022.
 DIRECT FOUR-DIMENSIONAL FLOW MEASUREMENT OF WSS IN THE THORACIC AORTA
Table 2. Peak systolic aWSS, cWSS, and OSI within the
aortic arch and descending aorta
Aortic Arch Descending Aorta
absWSS, Pa 1.79 ⫾ 0.71 2.23 ⫾ 1.04*
aWSS, Pa 1.50 ⫾ 0.74 2.13 ⫾ 1.08*
cWSS, Pa ⫺0.06 ⫾ 0.12 ⫺0.04 ⫾ 0.15
tWSS, Pa 0.87 ⫾ 0.31 0.90 ⫾ 0.37
OSI 0.07 ⫾ 0.04 0.06 ⫾ 0.05*
Values are means ⫾ SE. WSS, wall shear stress; aWSS, axial WSS; cWSS,
circumferential WSS; OSI, oscillatory shear index; absWSS, mean WSS
magnitude; tWSS, time-averaged WSS. Significance is shown for t-tests
comparing descending and aortic arch values. *P ⬍ 0.001.
in context with the results presented afterward. Statistical analyses
were performed using the statistical software package SPSS (IBM).
RESULTS
Clinical and demographic subject characteristics. A total of
224 subjects were included in the study from a total pool of 410
consecutive subjects recruited as part of the 4D-biomarker
study. In total, 111 patients were excluded. Individuals with
4D-flow data artifacts such as severe phase aliasing from
motion or poor gating were also excluded due to the unreliable
measurement of WSS in these cases (n ⫽ 75). The character-
istics of the cohort are shown in Table 1. Normal indexes of
cardiac function were present in ⬎95% of subjects.
Description of normative WSS values. Table 2 shows peak
systolic WSS and OSI measures in the aortic arch and descend-
ing aorta for the whole cohort. The average peak systolic WSS
magnitude value in the arch was 1.79 Pa across all subjects,
increasing to 2.23 Pa in the descending aorta (20% higher, P ⬍
0.0001). Axial WSS values showed a similar increase in the
descending aorta compared with the aortic arch (30% higher,
AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00017.2018 • www.ajpheart.org
Downloaded from journals.physiology.org/journal/ajpheart at Kangwon Natl Univ (203.252.078.033) on May 22, 2022.
H1177
P ⬍ 0.0001). No such difference between the arch and de-
scending aorta existed for circumferential WSS, tWSS, or OSI.
With the use of a regression model, the influence of age on
WSS magnitude was quantified as ⫺0.26 Pa/decade (R2 ⫽
0.405) in the aortic arch and ⫺0.36 Pa/decade (R2 ⫽ 0.378) in
the descending aorta (both P ⬍ 0.0001). Axial WSS measures
showed a similar trend, whereas tWSS displayed a much
attenuated and weaker decrease with age (aortic arch: ⫺0.06
Pa/decade, R2 ⫽ 0.127, P ⬍ 0.001; descending aorta: ⫺0.09
Pa/decade, R2 ⫽ 0.204, P ⬍ 0.001). Circumferential WSS and
OSI showed only a weak correlation to age (R2 ⬍ 0.03). No
influence of sex was detected for any WSS parameter.
Flat map representation of spatial and temporal variation in
WSS. Figure 3 shows flat maps of WSS magnitude (top), axial
WSS (middle), and circumferential WSS (bottom). The flat
map representation shows the aorta “unwrapped” and then
mapped to a rectangular distribution, with the top of the map
representing the aorta at the level of the pulmonary bifurcation
progressing distally along the thoracic aorta to the level of the
diaphragm at the bottom of the map (see Fig. 2). The distri-
bution WSS measures was highly heterogeneous, with WSS
magnitude and axial WSS showing a similar spatial pattern. A
localized region of high WSS magnitude and axial WSS was
present, originating on the anterior wall of the aortic arch and
extending longitudinally along the length of the thoracic aorta.
This feature occupied ~20% of the circumference of the aorta
in the early arch at systole, increasing in width at the distal
portion of the aortic arch.
Positive (clockwise orientated) and negative (anticlockwise
orientated) circumferential WSS zones were present in the
aortic arch. At high magnitude, positive region of circumfer-
ential WSS occurred at 6 o’clock in the proximal portion of the
aortic arch, and a negative region occurred between 11 and 1
Fig. 3. Parameterized representation of mean
wall shear stress (WSS) magnitude (ab-
sWSS), axial (aWSS), and circumferential
(cWSS) measurements by four-dimensional
flow in the aorta of 224 subjects. Measures
across the aorta have been parametrized to a
flat map representation and are displayed for
all of systole and early and late diastole. Mid
diastole (~30% of the cardiac cycle) is not
shown for conciseness.
H1178 DIRECT FOUR-DIMENSIONAL FLOW MEASUREMENT OF WSS IN THE THORACIC AORTA

o’clock, starting more distally in the arch. While the WSS and
axial WSS patterns reduced considerably during diastole, re-
sidual effects of the distinctive circumferential WSS patterns
remained throughout the entire cardiac cycle.
Time-averaged WSS and OSI. Figure 4 shows a flat map of
tWSS (effectively Fig. 3, top, averaged into a single image)
and OSI. The tWSS data reflect the same features observed in
both the WSS magnitude and axial WSS flat maps. OSI flat
maps showed low values in the proximal/mid descending aorta
(OSI ⬍ 0.15) representing the highly axial aortic flow in this
region (Fig. 4). High regions of OSI corresponded to locations
of likely retrograde flow, such as the proximal portion and crest
of the aortic arch. There was also evidence of disturbed flow
corresponding to disruption of axial flow as the distal descend-
ing aorta approached the diaphragm.
Variation of WSS patterns along the length of the aorta.
Figure 5A shows the variation in the mean axial WSS along the
length of the aorta at peak systole from all subjects. These data
reflect the general trend of lower values in the aortic arch
compared with descending aorta, with peak axial WSS values
in the mid descending aorta. Mean circumferential WSS
showed a distinctive dip at the distal portion of the aortic arch
representing high circumferential WSS of an anticlockwise
orientation (Fig. 5B). The variation of circumferential WSS
was lowest along the descending aorta, indicating a preference
for axially aligned flow in this region. During diastole, these
trends persisted for both axial WSS and circumferential WSS,
although at a lower magnitude of WSS values (data not
shown). Fig. 5. A: mean axial WSS (aWSS) along the length of the aorta at peak systole
from 224 normal aortas. B and C: corresponding plots for circumferential WSS
Mean OSI about the lumen circumference showed peak (cWSS; B) and oscillatory shear index (OSI; C) are also shown.
values of ~0.12 at the start of the aortic arch and at the level of
the diaphragm (Fig. 5C). The proximal descending portion of
the aorta had the lowest OSI values, indicating reduced back- “gothic” at 85 ⫾ 14° (where a perfectly round arch is 90°, and
flow compared with other regions. increasing triangular shape trending toward 0°), and aortic
Aortic shape and velocity measurements. The aortic form velocity was 53 ⫾ 18 cm/s.
and flow dynamics data are shown in Table 3. Mean aortic Modeling the associations of aortic form on WSS. A back-
length measured from the ascending aorta at the level of the ward-stepwise regression analysis was used to evaluate the
pulmonary bifurcation to the diaphragm across the cohort was influence of the aortic shape parameters on WSS. Adjusted R2
230 ⫾ 32 mm. Aortic diameter was 29 ⫾ 3 mm, mean radius values, standardized coefficients, and P values for each model
of curvature was 35 ⫾ 6 mm, average aortic shape was slightly are shown in Table 4. In the aortic arch, WSS magnitude had
a strong correlation (R2 ⫽ 0.58) with included variables vessel
diameter and velocity. Axial WSS in the aortic arch had a
strong correlation (R2 ⫽ 0.55) with velocity, AoCurve, and
AoShape. In the descending aorta again WSS magnitude and
axial WSS had strong correlations, whereas tWSS showed
good correlation (adjusted R2 ⫽ 0.37). For these three depen-
dent variables, vessel diameter, velocity, and AoCurve were
significant variables, whereas AoShape was not. Velocity was
the strongest predictor of WSS magnitude, axial WSS, and
tWSS models in both the arch and descending aorta.

Table 3. Aorta shape and hemodynamic factors.

Parameter Means ⫾ SD

AoShape 85.48 ⫾ 14.82

AoCurve, mm 34.71 ⫾ 6.12
Aortic arch diameter, mm 29.13 ⫾ 3.17
Aortic arch velocity, cm/s 45.74 ⫾ 13.31
Descending aortic diameter, mm 25.79 ⫾ 3.67
Descending aorta velocity, cm/s 52.51 ⫾ 18.42
Fig. 4. Time-averaged (tWSS) and oscillatory shear index (OSI) flat maps from
224 normal aortas. AoShape, aortic arch shape; AoCurve, aortic curve.

AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00017.2018 • www.ajpheart.org

Downloaded from journals.physiology.org/journal/ajpheart at Kangwon Natl Univ (203.252.078.033) on May 22, 2022.
 DIRECT FOUR-DIMENSIONAL FLOW MEASUREMENT OF WSS IN THE THORACIC AORTA H1179
Table 4. Influence of aortic geometry and velocity parameters on WSS in the aortic arch and descending aorta: summary of
regression analysis.
AoShape AoCurve Diameter Velocity

Adjusted R2 ␤ P ␤ P ␤ P ␤ P

Aortic arch
absWSS 0.58 ⫺0.10 0.03 0.73 ⬍0.001
aWSS 0.55 ⫺0.29 0.003 0.21 0.03 0.70 ⬍0.001
cWSS 0.03 0.18 0.01
tWSS 0.18 ⫺0.13 0.04 0.38 ⬍0.001
OSI 0.07 0.15 0.05 ⫺0.13 0.06 0.25 ⬍0.001
Descending aorta
absWSS 0.66 0.17 0.001 ⫺0.21 ⬍0.001 0.75 ⬍0.001
aWSS 0.62 0.18 0.001 ⫺0.23 ⬍0.001 0.72 ⬍0.001
cWSS 0.07 ⫺0.49 0.001 0.45 0.001 ⫺0.22 0.004
tWSS 0.37 0.19 0.01 ⫺0.17 0.02 0.58 ⬍0.001
OSI 0.02 ⫺0.24 0.01 0.17 0.05
WSS, wall shear stress; AoShape, aortic arch shape; AoCurve, aortic curve; absWSS, mean WSS magnitude; aWSS, axial WSS; cWSS, circumferential WSS;
tWSS, time-averaged WSS; OSI, oscillatory shear index. For each model, adjusted R2 values and standardized coefficients (␤) and P values of the included
variables are shown.

DISCUSSION there has been no clear technique capable of accurately mea-

This study provides the most complete description of the
normal range of thoracic aorta WSS available. Our rigorous
approach permits quantification of the influence of key geo-
metric and hemodynamic parameters on the spatial and tem-
poral patterns of WSS in the thoracic aorta. We used an
optimized multi-VENC 4D-flow MRI approach, including
novel vector field cleaning and postprocessing methods to
overcome previous difficulties in obtaining direct measure-
ments of WSS. Our key findings are that WSS patterns are
highly geographically variable, with peak values characteristi-
cally occurring on the anterior wall in the mid descending
aorta. WSS values have a strong age dependence and are
predominantly governed by velocity, vessel diameter, and
radius of curvature of the aortic arch.
No previous study has directly measured WSS in the aortic
arch across a large population of subjects with no known aortic
pathology and normal left ventricular function. Frydrychowicz
et al. (17) previously presented 4D-flow derived aortic WSS in
a cohort of 31 subjects. A number of general trends of their
data agree with the results presented here, such as higher WSS
in the descending aorta compared with aortic arch and the
presence of positive circumferential WSS in the aortic arch,
becoming negative in the descending aorta. However, their
WSS values were much lower, likely attributed to measure-
ment technique, and were measured at sparsely spaced cross
sections. It is well known that the relationship of hemodynamic
and shape factors is important in governing the secondary flow
dynamics of the aorta (14, 15); however, there is little in the
way of explicit demonstration of the quantitative nature of
these relationships. Our data demonstrate an influence of cur-
vature and shape on WSS values, even with the relatively tight
distribution of these parameters in our cohort with normal
aortic anatomy and no known pathology. Additionally, WSS
was shown to strongly decrease with age. The data presented
will provide a valuable reference range of “optimal” WSS
magnitude and patterns under ideal flow conditions, enabling a
greater understanding of hemodynamic induced vessel wall
disease.
While previous studies have identified a clear relationship
between WSS and vascular remodeling (26, 30), until recently
suring WSS in vivo or providing a reliable description of a
normal population dynamics. In fact, conflicting information
from low number studies and imperfect measurement tech-
niques have led to confusion and contradiction in the field (21).
Similarly, evidence for a causative role for WSS in the patho-
biology of aortic dilatation largely rests on ex vivo studies or
from settings of abnormal hemodynamics such as animal
models or patients with bicuspid valve (5, 30). We believe that
this study provides much-needed clarification of what the
physiological WSS looks like in a normal population and how
the geometry and flow dynamics of the aorta influence this.
The method showcased by this report would be an ideal means
to collect further direct evidence to evaluate the explicit effects
of regional WSS in these conditions. Additional sources of
stress on the aortic wall, such as circumferential and longitu-
dinal stresses, exist in response to pressure pulsation and heart
and diaphragm motion. These stresses are many magnitudes
greater than the WSS discussed here; however, it is generally
accepted that WSS drives the signaling for pathological vas-
cular remodeling (13). Future work aimed at understanding the
relationships between anatomy and other physiological influ-
ences on WSS will likely need to apply innovative methods as
this is a highly nonlinear problem, e.g., Dasi et al. (12).
4D-flow MRI has some important limitations that affect
WSS measurement. These are largely a consequence of the low
spatial resolution, relative to the near wall velocity gradients,
limits on vessel wall segmentation accuracy; errors in estima-
tion secondary to wall movement, and current technical limi-
tations on the accuracy of velocity measurement (25). In this
work, we have adopted several steps to overcome these limi-
tations. Using a multi-VENC encoding strategy, we maximize
the vector-to-noise ratio. Applying the previously described
VFIT postprocessing technique is an effective means of de-
creasing noise and increasing the spatial resolution. We also
limited our area of interest to the aortic lumen beginning at the
level of the main pulmonary artery bifurcation to eliminate
inaccuracies induced by the movement of the ascending aorta.
Previous applications of 4D-flow MRI to WSS measurement
have recognized an underestimation of WSS magnitudes with
respect to data published in numerical studies, although have

AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00017.2018 • www.ajpheart.org

Downloaded from journals.physiology.org/journal/ajpheart at Kangwon Natl Univ (203.252.078.033) on May 22, 2022.
H1180 DIRECT FOUR-DIMENSIONAL FLOW MEASUREMENT OF WSS IN THE THORACIC AORTA
showed good reproducibility and low interobserver variability
(31). We have previously demonstrated that WSS derived from
the VFIT technique compares favorably with data from com-
putational studies (6). Consequently, our WSS magnitude val-
ues are approximately 5–10 times larger than previously re-
ported 4D-flow MRI WSS values and are in close in agreement
with simulation data (3, 18). While broad agreement with CFD
measurements is useful confirmatory evidence that our data are
realistic, it would be incorrect to propose CFD as a “gold
standard” for WSS since there are many assumptions inherent
in this method (e.g., boundary condition definition and gov-
erning equation linearization). Current work in our laboratory
and elsewhere is aimed at combination approaches using mea-
sured data and CFD to overcome these limitations of CFD and
the inherent limits of current direct measurements (7). Further
steps are required to establish the current approach as a viable
method in the clinic including evaluation of the test-retest
stability of the measurements, replication in additional normal
and diseased cohorts, and cross-platform availability of the
method. MRI techniques such as Fourier velocity encoding are
advanced methods that allow improved WSS estimation but at
a cost of greater imaging and postprocessing complexity (25).
WSS measurements are directly proportional to viscosity,
which we have assumed constant for all subjects in this study.
Further work is necessary to investigate the influence of vis-
cosity variation that may exist due to aging, shear rate or blood
disorders.
Conclusions. Our results demonstrate that the normal range
of thoracic WSS depends on location, follows a trend of
reduction with age, and is strongly correlated with velocity,
vessel diameter, and aortic arch radius of curvature. To enable
analysis of WSS over a large population, we applied a param-
eterization technique that presented data in a regular, flat map
fashion suitable for group or individual analysis. The approach
described here is suitable for widespread application; however,
translation into the clinic would require broader application to
diseased cohorts and detailed evaluation of WSS as a bio-
marker in outcome studies.
GRANTS
This work was supported by the Heart Research Institute, Sydney Medical
School Foundation, HeartKids and Parker Hughes Bequest, and University of
Sydney (Sydney, NSW, Australia).
DISCLOSURES
No conflicts of interest, financial or otherwise, are declared by the authors.
AUTHOR CONTRIBUTIONS
F.M.C. and S.M.G. conceived and designed research; F.M.C. and S.M.G.
performed experiments; F.M.C. and S.M.G. analyzed data; F.M.C. and S.M.G.
interpreted results of experiments; F.M.C. and S.M.G. prepared figures; F.M.C.
and S.M.G. drafted manuscript; F.M.C. and S.M.G. edited and revised man-
uscript; F.M.C. and S.M.G. approved final version of manuscript.
REFERENCES
1. Antiga L, Piccinelli M, Botti L, Ene-Iordache B, Remuzzi A, Steinman
DA. An image-based modeling framework for patient-specific computa-
tional hemodynamics. Med Biol Eng Comput 46: 1097–1112, 2008.
doi:10.1007/s11517-008-0420-1.
2. Arvidsson PM, Töger J, Heiberg E, Carlsson M, Arheden H. Quanti-
fication of left and right atrial kinetic energy using four-dimensional
intracardiac magnetic resonance imaging flow measurements. J Appl
Physiol 114: 1472–1481, 2013. doi:10.1152/japplphysiol.00932.2012.
AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00017.2018 • www.ajpheart.org
Downloaded from journals.physiology.org/journal/ajpheart at Kangwon Natl Univ (203.252.078.033) on May 22, 2022.
3. Biasetti J, Hussain F, Gasser TC. Blood flow and coherent vortices in
the normal and aneurysmatic aortas: a fluid dynamical approach to
intra-luminal thrombus formation. J R Soc Interface 8: 1449 –1461, 2011.
doi:10.1098/rsif.2011.0041.
4. Boussel L, Rayz V, McCulloch C, Martin A, Acevedo-Bolton G,
Lawton M, Higashida R, Smith WS, Young WL, Saloner D. Aneurysm
growth occurs at region of low wall shear stress: patient-specific correla-
tion of hemodynamics and growth in a longitudinal study. Stroke 39:
2997–3002, 2008. doi:10.1161/STROKEAHA.108.521617.
5. Bürk J, Blanke P, Stankovic Z, Barker A, Russe M, Geiger J,
Frydrychowicz A, Langer M, Markl M. Evaluation of 3D blood flow
patterns and wall shear stress in the normal and dilated thoracic aorta using
flow-sensitive 4D CMR. J Cardiovasc Magn Reson 14: 84, 2012. doi:10.
1186/1532-429X-14-84.
6. Callaghan FM, Grieve SM. Spatial resolution and velocity field improve-
ment of 4D-flow MRI. Magn Reson Med 78: 1959 –1968, 2017. doi:10.
1002/mrm.26557.
7. Callaghan FM, Karkouri J, Broadhouse K, Evin M, Fletcher DF,
Grieve SM. Thoracic aortic aneurysm: 4D flow MRI and computational
fluid dynamics model. Comput Methods Biomech Biomed Engin 18, Suppl
1: 1894 –1895, 2015. doi:10.1080/10255842.2015.1069559.
8. Callaghan FM, Kozor R, Sherrah AG, Vallely M, Celermajer D,
Figtree GA, Grieve SM. Use of multi-velocity encoding 4D flow MRI to
improve quantification of flow patterns in the aorta. J Magn Reson
Imaging 43: 352–363, 2016. doi:10.1002/jmri.24991.
9. Carlhäll CJ, Bolger A. Passing strange: flow in the failing ventricle. Circ
Heart Fail 3: 326 –331, 2010. doi:10.1161/CIRCHEARTFAILURE.109.
911867.
10. Cibis M, Potters WV, Gijsen FJ, Marquering H, vanBavel E, van der
Steen AF, Nederveen AJ, Wentzel JJ. Wall shear stress calculations
based on 3D cine phase contrast MRI and computational fluid dynamics:
a comparison study in healthy carotid arteries. NMR Biomed 27: 826 –834,
2014. doi:10.1002/nbm.3126.
11. Craiem D, Chironi G, Casciaro ME, Graf S, Simon A. Calcifications of
the thoracic aorta on extended non-contrast-enhanced cardiac CT. PLoS
One 9: e109584, 2014. doi:10.1371/journal.pone.0109584.
12. Dasi LP, Pekkan K, Katajima HD, Yoganathan AP. Functional analysis
of Fontan energy dissipation. J Biomech 41: 2246 –2252, 2008. doi:10.
1016/j.jbiomech.2008.04.011.
13. Davies PF. Flow-mediated endothelial mechanotransduction. Physiol Rev
75: 519 –560, 1995. doi:10.1152/physrev.1995.75.3.519.
14. Dean WR. LXXII. The stream-line motion of fluid in a curved pipe
(second paper). Lond Edinb Dublin Philos Mag J Sci 5: 673–695, 1928.
doi:10.1080/14786440408564513.
15. Dean WR. XVI. Note on the motion of fluid in a curved pipe. Lond
Edinb Dublin Philos Mag J Sci 4: 208 –223, 1927. doi:10.1080/
14786440708564324.
16. Dyverfeldt P, Bissell M, Barker AJ, Bolger AF, Carlhäll CJ, Ebbers T,
Francios CJ, Frydrychowicz A, Geiger J, Giese D, Hope MD, Kilner
PJ, Kozerke S, Myerson S, Neubauer S, Wieben O, Markl M. 4D flow
cardiovascular magnetic resonance consensus statement. J Cardiovasc
Magn Reson 17: 72, 2015. doi:10.1186/s12968-015-0174-5.
17. Frydrychowicz A, Stalder AF, Russe MF, Bock J, Bauer S, Harloff A,
Berger A, Langer M, Hennig J, Markl M. Three-dimensional analysis
of segmental wall shear stress in the aorta by flow-sensitive four-dimen-
sional-MRI. J Magn Reson Imaging 30: 77–84, 2009. doi:10.1002/jmri.
21790.
18. Karmonik C, Müller-Eschner M, Partovi S, Geisbüsch P, Ganten MK,
Bismuth J, Davies MG, Böckler D, Loebe M, Lumsden AB, von
Tengg-Kobligk H. Computational fluid dynamics investigation of chronic
aortic dissection hemodynamics versus normal aorta. Vasc Endovascular
Surg 47: 625–631, 2013. doi:10.1177/1538574413503561.
19. Li YS, Haga JH, Chien S. Molecular basis of the effects of shear stress
on vascular endothelial cells. J Biomech 38: 1949 –1971, 2005. doi:10.
1016/j.jbiomech.2004.09.030.
20. Markl M, Brendecke SM, Simon J, Barker AJ, Weiller C, Harloff A.
Co-registration of the distribution of wall shear stress and 140 complex
plaques of the aorta. Magn Reson Imaging 31: 1156 –1162, 2013. doi:10.
1016/j.mri.2013.05.001.
21. Meng H, Tutino VM, Xiang J, Siddiqui A. High WSS or low WSS?
Complex interactions of hemodynamics with intracranial aneurysm initi-
ation, growth, and rupture: toward a unifying hypothesis. AJNR Am J
Neuroradiol 35: 1254 –1262, 2014. doi:10.3174/ajnr.A3558.
 DIRECT FOUR-DIMENSIONAL FLOW MEASUREMENT OF WSS IN THE THORACIC AORTA
22. Metaxa E, Tremmel M, Natarajan SK, Xiang J, Paluch RA, Mandel-
baum M, Siddiqui AH, Kolega J, Mocco J, Meng H. Characterization of
critical hemodynamics contributing to aneurysmal remodeling at the
basilar terminus in a rabbit model. Stroke 41: 1774 –1782, 2010. doi:10.
1161/STROKEAHA.110.585992.
23. Peiffer V, Sherwin SJ, Weinberg PD. Does low and oscillatory wall
shear stress correlate spatially with early atherosclerosis? A systematic
review. Cardiovasc Res 99: 242–250, 2013. doi:10.1093/cvr/cvt044.
24. Pelc NJ, Bernstein MA, Shimakawa A, Glover GH. Encoding strategies
for three-direction phase-contrast MR imaging of flow. J Magn Reson
Imaging 1: 405–413, 1991. doi:10.1002/jmri.1880010404.
25. Petersson S, Dyverfeldt P, Ebbers T. Assessment of the accuracy of
MRI wall shear stress estimation using numerical simulations. J Magn
Reson Imaging 36: 128 –138, 2012. doi:10.1002/jmri.23610.
26. Raut SS, Chandra S, Shum J, Finol EA. The role of geometric and
biomechanical factors in abdominal aortic aneurysm rupture risk assess-
ment. Ann Biomed Eng 41: 1459 –1477, 2013. doi:10.1007/s10439-013-
0786-6.
27. Roberts WC. Aortic dissection: anatomy, consequences, and causes. Am
Heart J 101: 195–214, 1981. doi:10.1016/0002-8703(81)90666-9.
28. Rogers IS, Massaro JM, Truong QA, Mahabadi AA, Kriegel MF, Fox
CS, Thanassoulis G, Isselbacher EM, Hoffmann U, O’Donnell CJ.
H1181
Distribution, determinants, and normal reference values of thoracic and
abdominal aortic diameters by computed tomography (from the Framing-
ham Heart Study). Am J Cardiol 111: 1510 –1516, 2013. doi:10.1016/j.
amjcard.2013.01.306.
29. Schroeder W, Martin K, Lorensen B. The Visualization Toolkit (3rd
ed.). New York: Kitware, 2004.
30. van Ooij P, Potters WV, Collins J, Carr M, Carr J, Malaisrie SC,
Fedak PW, McCarthy PM, Markl M, Barker AJ. Characterization of
abnormal wall shear stress using 4D flow MRI in human bicuspid
aortopathy. Ann Biomed Eng 43: 1385–1397, 2015. doi:10.1007/s10439-
014-1092-7.
31. van Ooij P, Powell AL, Potters WV, Carr JC, Markl M, Barker AJ.
Reproducibility and interobserver variability of systolic blood flow
velocity and 3D wall shear stress derived from 4D flow MRI in the
healthy aorta. J Magn Reson Imaging 43: 236 –248, 2016. doi:10.1002/
jmri.24959.
32. von Allmen RS, Anjum A, Powell JT. Incidence of descending aortic
pathology and evaluation of the impact of thoracic endovascular aortic
repair: a population-based study in England and Wales from 1999 to 2010.
Eur J Vasc Endovasc Surg 45: 154 –159, 2013. doi:10.1016/j.ejvs.2012.
12.007.

AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00017.2018 • www.ajpheart.org

Downloaded from journals.physiology.org/journal/ajpheart at Kangwon Natl Univ (203.252.078.033) on May 22, 2022.