Key words
angiography - velocity ratio - atherosclerosis - peripheral arterial disease - vector
velocity ultrasound
Introduction
Atherosclerosis is a systemic disease of the larger arteries causing luminal narrowing
leading to cardiovascular diseases, the number one cause of death globally [1]. One subgroup of cardiovascular diseases is peripheral arterial disease (PAD), in
particular lower extremity arterial disease.
Digital subtraction angiography (DSA) is the gold standard for vessel analysis. Besides
a diagnostic overview of the blood vessels, it allows simultaneous endovascular therapy.
However, DSA is invasive, contrast-dependent, and exposes patients and staff to ionizing
radiation [2]. Moreover, traditional DSA is a two-dimensional imaging modality, limiting the stenosis
assessment to the imaged projection plane. This can lead to inaccurate diameter measurements,
especially for elliptic and other complicated vessel shapes.
An alternative to DSA is Doppler ultrasound, which is a diagnostic tool providing
information regarding the vessel wall and hemodynamics and is commonly used for evaluation
of PAD. Color and spectral Doppler can be used to localize and assess the severity
of possible stenoses by estimating the blood flow velocity. The velocity correlates
with lumen diameter, but due to interindividual variations of blood flow, velocity
ratios, i. e. the intrastenotic velocity divided by the prestenotic velocity, provide
better estimates of the stenoses [3]
[4]. Velocity ratios are normally calculated on the basis of two individual peak systolic
velocity measurements, measured in the stenosis and in a proximal vessel segment with
a normal lumen. In previous studies, velocity ratios varying from 1.5 to 2.4 have
been shown to distinguish<50% stenoses from>50% stenoses, where the stenosis degree
was based on angiographic diameter reduction [3]
[4]
[5]
[6]
[7]. The conventional Doppler technique is angle-dependent and operator-reliant, in
particular in the presence of stenoses [8]
[9]. Operator errors can therefore lead to substantial deviations in velocity estimation
and subsequent velocity ratio calculation, and because two individual velocity estimations
are needed to provide a ratio, the deviation can be further aggravated.
To circumvent the angle dependency of conventional Doppler, the angle-independent
ultrasound technique Vector Flow Imaging (VFI) was proposed by Jensen and Munk [10]. VFI provides simultaneously the axial and transverse velocity components of the
blood flow. A conventional ultrasound pulse for flow estimation is transmitted, and
the received echoes are beamformed to yield three beams in parallel. One uses conventional
beamforming for estimating the axial velocity, and the other two beams are used for
estimating the transverse velocity component. By combining the velocity components
along the two axes, 2D vector velocities are obtained. VFI using linear arrays has
a tissue penetration of 5 cm and is useful on superficial blood vessels. The technique
is described further by [10]
[11]
[12], and the clinical use by [13]
[14]
[15]
[16]
[17]
[18]
[19].
The aim of the study was to investigate VFI as a technique for the quantitative assessment
of PAD. The technique was tested in a small patient group with PAD and clinical indication
of stenosis in the superficial femoral artery (SFA). The hypothesis was that velocity
ratios derived from VFI can be used to distinguish significant stenoses (>50% diameter
reduction) from non-significant stenoses. DSA was used as the reference technique
to measure vessel diameter.
Materials and Methods
Patients
Thirty consecutive patients scheduled for DSA of the lower extremities due to suspected
PAD were examined. Patients were eligible for inclusion if they had one or more previously
untreated atherosclerotic lesions (stenosis or plaque) in the SFA. Nineteen patients
with previous by-pass surgery, endovascular surgery, occlusion, no lesions (judged
by both ultrasound and DSA), or widespread atherosclerotic disease according to the
TransAtlantic Inter-Society Consensus Document on Management of PAD (TASC II, [20]) were excluded. 11 of the 30 patients were included, providing a total of 16 lesions
consisting of 13 stenoses and 3 plaques.
Written informed consent was obtained. The local Ethics Committee waived approval,
since ultrasound scanning of atherosclerotic extremities is considered a routine procedure
(protocol number: H-4-2013-001).
Scan setup using Vector Flow Imaging
A commercial scanner (UltraView 800, BK Ultrasound, Herlev, Denmark) was used with
a linear transducer with a center frequency of 9 MHz (8670, BK Ultrasound, Herlev,
Denmark).
All patients underwent ultrasound scan in the angiography room just prior to DSA,
and all were scanned in a supine position after at least 15 min. of rest. The patients'
SFAs were scanned longitudinally from the bifurcation of the common femoral artery
to the point where the SFA enters the adductor canal. When disturbed flow was detected
by VFI, a marker (paper clip) was attached to the patient's thigh corresponding to
the location of the flow disturbance ensuring corresponding ultrasound and angiographic
recordings ([Fig. 1]). From this location a VFI-sequence of 15 s was recorded with a frame rate of 15 Hz.
The recording contained flow both in the lesion and proximal/distal to the lesion.
Disturbed flow was defined as the presence of vortices, flow in multiple directions
and/or suddenly occurring aliasing indicating increasing flow velocities. VFI provides
2D images of the blood flow, where each pixel contains quantitative information about
direction and velocity with superimposed vector arrows to facilitate flow visualization
([Fig. 2]).
Fig. 1 DSA with paper clip marker indicating the stenosis (patient 7).
Fig. 2 Scanning of a stenosis (patient 4) using VFI. The arrows illustrate flow direction
and relative velocity magnitude. The arrows are for illustrative purposes only and
are not used for the quantitative estimation of velocity and direction. The blood
flows from left to right as indicated by the arrows in the green area. Aliasing indicating
higher flow velocities is seen in the purple area to the left and poststenotic disturbed
flow is seen to the right. Notice the angle of insonation of 90º.
The color box is operated similar to color Doppler and was adjusted to cover the vessel
and vessel walls. The pulse repetition frequency (PRF) was adjusted to the level providing
the best possible filling of the vessel, even if aliasing still was present in peak
systole. If the PRF was adjusted to the level where no aliasing was present, data
containing lower flow velocities would be neglected. The angle of insonation was 70-90
degrees in all cases.
Previous VFI studies performed on a flow rig and in-vivo indicated a negative bias
around -10% for velocity estimations [12]
[13]. This is due to a bias in the estimation scheme, which can be compensated for in
an optimized setup as demonstrated by Jensen et al. [21]. However, in this study determination of the exact velocities is not relevant, since
the velocity estimations are used to calculate velocity ratios, and thus, any systematic
error is removed.
Angiography
An Infinix-i system (model INFX-8000 V, Toshiba Medical Systems Corporation, Tochigi-ken,
Japan) was used for DSA. Puncture of the common femoral artery was performed followed
by placement of a 5 French sheath. A 4 or 5 French catheter was used for contrast
injections and DSA was performed using 2 frames/s and a 6-10 ml. contrast injection
(Visipaque 270 mgI/ml). Routine anteroposterior images in one plane were recorded
and occasionally supplemented by oblique projections. Subsequent measurements were
performed on a standard workstation. From the region of interest (marked with the
paper clip), the image yielding the most severe diameter reduction was used for calculation
of the stenosis degree percentage. This was calculated using the smallest diameter
in the stenosis versus the diameter in an adjacent normal arterial segment. A stenosis
degree of 40% corresponds to a vessel diameter reduced by 40% compared to the normal
vessel. Disturbed flow detected with VFI with no corresponding diameter reduction
in any of the DSA images was defined as an atherosclerotic plaque. Stenosis degree
percentage was calculated independently of the ultrasound scanning by a radiologist
not otherwise involved in the study.
Velocity ratios calculated from VFI
The VFI recordings were analyzed off-line with in-house MATLAB scripts (MathWorks,
Natick, MA, USA) with a point-and-click interface providing the velocity and direction
for each pixel. From each recording, three frames illustrating flow with the best
possible filling of the vessel in both the lesion and healthy part of the SFA were
selected. The velocity ratio was then calculated as the maximum velocity detected
as centrally as possible in the lesion divided by the maximum velocity detected centrally
in the adjacent disease-free segment. In three cases the challenge was to obtain velocities
upstream of the stenosis, and thus, only downstream velocities were measured. No measurements
were made in the turbulence immediately downstream of the stenosis. Both velocities
were obtained from the same frame, and as far from each other as possible within the
width of the transducer. Only velocities with a flow direction parallel to the vessel
wall were used, excluding velocities in regions of turbulence. The maximum velocities
were located manually in each selected frame from the colored pixels of VFI via the
point-and-click interface ([Fig. 3]). The final velocity ratio was calculated as the average of the velocity ratios
from the three frames. If shadowing from a calcified plaque in the superficial vessel
wall was present, maximum velocities were obtained from either side of the shadow,
and the velocity obtained distal to the lesion was divided with the velocity obtained
proximal to the lesion.
Fig. 3 The top image shows the MATLAB processed VFI recording of the stenosis illustrated
by the DSA in the lower image (patient 3). The top image represents the part of the
vessel shown in the blue box in the lower image with the clinically relevant stenosis
in the middle. The color bar to the right of the top image shows the velocity range
in cm/s for this specific frame. The color bar is not used for quantitative estimation,
only for orientation. The blood flows from left to right. Maximum velocities around
25 cm/s are detected in the red area and in the turquoise area to the right of the
stenosis velocities around 7 cm/s are detected. The yellow area immediately poststenotic
represents the flow jet with a velocity of 13-14 cm/s. These velocities are obtained
from a random point in the cardiac cycle where the best possible filling of the vessel
is seen without aliasing being present. The marker is not visible in this projection.
Assuming a high level of arterial stiffness due to atherosclerosis throughout the
SFA, the relationship between the cross-sectional areas in the diseased versus the
non-diseased vessel segments remains constant during the heart cycle. The relationship
between the velocities, i. e. the velocity ratio, remains constant too, when the obtained
velocities are from the same point of the cardiac cycle. With a pulse wave velocity
of at least 10 m/s [22]
[23] and a 4 cm linear transducer the region of the vessel covered by the transducer
is passed in maximum 4 ms. With a frame rate of 15 Hz it is therefore assumed that
velocities obtained within the same frame are from the same point of the cardiac cycle.
All velocity ratios were calculated blinded to the estimation of angiographic stenosis
degree.
Statistical analysis
The correlation of average velocity ratios and angiographic stenosis degree was estimated
by nonlinear regression analysis. The velocity ratio corresponding to a 50% stenosis
was calculated. Stenoses>50% and<50% were treated as two different groups, and mean
velocity ratios including standard deviations were calculated for each group and compared.
An unpaired t-test was performed and p<0.05 was considered significant. All calculations were performed both with and without
outliers. To further assess the ability of the velocity ratios to predict stenoses>50%,
ROC analyses were performed and the AUC was reported. All statistical calculations
were performed in LibreOffice Calc except ROC analysis, which was done in R (version
3.3.1).
Results
All three calculated velocity ratios, the average velocity ratio, and the angiographic
stenosis degree (expressed as percentage reduction) of each lesion are shown in [Table 1]. Two lesions are considered outliers. Patient 8 has a velocity ratio of 2.5 and
a stenosis degree of 11%, and patient 11 has a velocity ratio of 1 and a stenosis
degree of 67%. The correlation between the average velocity ratios and stenosis degrees
is illustrated in [Fig. 4] with and without the outliers. Without the outliers, the velocity ratio corresponding
to a 50% stenosis is 2.1, and with all lesions included the velocity ratio is 2.5.
Fig. 4 Correlation between average velocity ratios and angiographic diameter reduction expressed
as stenosis percentage. The correlation has been illustrated for all data (top) and
with the two outliers omitted (bottom). Notice that the ideal correlation line (assuming
parabolic flow) starts in (1, 0) with a velocity ratio of 1 when no stenosis is present.
The regression lines were generated automatically by LibreOffice Calc.
Table 1 Velocity ratios and corresponding stenosis degrees.
|
Patient number
|
Lesion number
|
Lesion type
|
Velocity ratios
|
Average velocity ratio
|
Degree of stenosis (%)
|
|
1
|
1
|
Stenosis
|
2.1, 1.9, 2.7
|
2.2
|
78
|
|
2
|
Plaque
|
1.1, 0.9, 1.2
|
1.1
|
0
|
|
2
|
1
|
Plaque
|
0.9, 1, 1
|
1
|
0
|
|
2
|
Stenosis
|
1.2, 1.2, 1.3
|
1.2
|
19
|
|
3
|
1
|
Stenosis
|
2.6, 3.6, 2.6
|
2.9
|
68
|
|
4
|
1
|
Stenosis
|
1.6, 4.4, 1.7
|
2.6
|
65
|
|
5
|
1
|
Stenosis
|
1.2, 1, 1.3
|
1.2
|
37
|
|
2
|
Stenosis
|
0.7, 0.8, 1.3
|
0.9
|
31
|
|
6
|
1
|
Stenosis
|
2.4, 1.8, 2.1
|
2.1
|
33
|
|
2
|
Stenosis
|
1.3, 1.6, 1.5
|
1.5
|
15
|
|
3
|
Stenosis
|
1, 1.3, 1.2
|
1.2
|
15
|
|
7
|
1
|
Stenosis
|
2.1, 2.4, 2.1
|
2.2
|
62
|
|
8
|
1
|
Stenosis
|
1.9, 2.9, 2.8
|
2.5
|
11
|
|
9
|
1
|
Stenosis
|
1.2, 1, 1.1
|
1.1
|
47
|
|
10
|
1
|
Plaque
|
1.3, 1.3, 1.2
|
1.3
|
0
|
|
11
|
1
|
Stenosis
|
1, 1.1, 1
|
1
|
67
|
Velocity ratios based on VFI recordings from each individual lesion and coherent stenosis
degree based on angiographic diameter reduction. A plaque is defined as a flow disturbing
lesion with no corresponding angiographic diameter reduction.
With patient 8 and patient 11 excluded from the analysis, the mean velocity ratio
(based on the average velocity ratios) for an angiographic stenosis degree<50% is
1.3 (standard deviation (SD) 0.34) and the mean velocity ratio for an angiographic
stenosis degree>50% is 2.5 (SD 0.34). The difference between the two groups is significant
(p<0.01). With all patients included in the analysis, the mean velocity ratios for stenosis
degrees<50% and>50% are 1.4 (SD 0.49) and 2.2 (SD 0.72), respectively. The difference
is still significant (p=0.02). Based on all patients, the AUC was 0.79 (95% CI: 0.46 to 1) and excluding
the two outliers the AUC was 0.95 (95% CI: 0.84 to 1).
In patients 3 and 5 (both lesions), the velocity ratios are based on downstream poststenotic
velocities, but no significant differences separate them from the remaining ratios
based on upstream velocities.
In one case (patient 1, lesion 1) the paper clip marker pointed towards a point approximately
2 cm from the stenosis, and in the remaining cases it pointed directly towards the
stenosis.
Discussion
To the authors' knowledge, this study is the first to grade arterial stenoses in the
SFA using vector velocity ultrasound. Even though the patient number is small, the
obtained velocity ratios of 2.1 (without outliers) and 2.5 (all data) corresponding
to a 50% stenosis match previous larger studies based on spectral Doppler, and the
difference between the two patient groups (<50% and>50% stenosis) is statistically
significant, both with and without the outliers. This is further supported by the
ROC analysis. The hypothesis of the study is therefore accepted.
Use of VFI is more intuitive than conventional Doppler as angle-independent velocities
are provided, thereby making considerations of insonation angle and angle correction
unnecessary. Moreover, VFI provides quantitative blood flow information for the full
vector map, thus more flow data are available to assess flow changes. This can potentially
help physicians diagnose PAD more effectively, spare patients unnecessary examinations,
and save time in daily clinical practice. Also, it is not necessary to assume where
in the stenotic vessel the peak velocities are found as it is with spectral Doppler
when positioning the range gate, as all detected velocities are given within the vector
map. However, at this stage the off-line analysis of the VFI recordings described
previously is very time-consuming and takes 60–90 min for each lesion. Further development
of VFI and implementation of scripts providing all velocities in real-time on the
scanner is therefore necessary, before the full potential is exploited and daily clinical
use is realistic.
The major limitations of this study are the small patient number and the lack of comparison
to conventional spectral Doppler. VFI was not compared to spectral Doppler as the
latter is not used in clinical preparations before referral to endovascular interventions
at our unit. However, in a future larger scale study, both spectral Doppler measurements
and DSA should be regarded as reference standards.
The use of VFI in this study is limited by the manual acquisition of all velocities
via the point-and-click interface in the MATLAB scripts. Subsequently, the number
of velocity measurements used for each velocity ratio calculation was challenged and
vulnerable for erroneous measurements, which may explain outlier patient 11 and the
wide range of velocity ratios for patient 4 reported in [Table 1]. Thus, the process should be automated in future studies.
VFI is dependent on the PRF, and in a stenotic artery with major velocity fluctuations
(the velocity increases more than five times when a stenosis exceeds 80% [3]), numerous frames will be affected by either aliasing, when the PRF is set too low,
or no velocity information, when the PRF is set too high. This further limits the
number of usable frames from each VFI recording.
To ensure that the velocities measured with VFI are from the same point in the cardiac
cycle, the velocities must be obtained within the same frame, limiting the size of
the region of interest to the width of the transducer. Another limitation of ultrasonic
grading of stenoses, whether VFI or spectral Doppler, is the 2D visualization of the
region of interest, which can affect the positioning of the transducer relative to
the artery, and hence the velocity estimation [24]. Also, the velocity ratios were calculated by one radiologist only, and inter-observer
variation was therefore not found.
The presence of calcified plaques in the vessel wall, and difficult visualization
of the small lumen in a severely atherosclerotic vessel provide other ultrasonic challenges.
The region of interest can be covered by a shadow, or the maximum velocities are not
visualized exactly where expected, e. g. in [Fig. 3] where peak velocities apparently are detected immediately proximal to the stenosis
and not in the stenosis. A possible explanation can be that calcified plaques in the
vessel wall disturb the flow signal in the stenotic segment, or the flow can be eccentric
in the stenosis and therefore out-of-plane as described for aortic stenosis [25]
[26]
[27]. Another reason could be that the entire stenosis is not visualized sufficiently
by DSA because of the given anteroposterior image plane. Hence, the stenosis could
actually also cover the vessel segment corresponding to the red area of the ultrasound
recording.
DSA is the gold standard for diagnosing and grading PAD, but is, just as ultrasound,
a 2D visualization of the vessels, and underestimation of stenoses can therefore occur
if the smallest diameter of the vessel is not visible in the angiographic projection.
DSA is occasionally supplemented by oblique projections if doubt about a stenosis
is raised, but that is no guarantee for a projection illustrating the most severe
stenosis. Angiographic underestimation of the stenosis could explain outlier patient
8.
In conclusion, this study has for the first time characterized atherosclerotic stenoses
and plaques in the SFA using velocity ratios obtained with a commercially available
vector velocity ultrasound technique. A velocity ratio of 2.5 has been shown to distinguish
between stenoses over and under 50% angiographic diameter reduction, and patients
with clinically relevant stenoses>50% have been identified with statistical significance.
The technique has potential to be used for monitoring atherosclerotic patients and
to support the indication of referral to DSA by pointing out and grading potential
stenoses in advance, thereby avoiding unnecessary angiographies.
