Content uploaded by Christian Kollmann
Author content
All content in this area was uploaded by Christian Kollmann on Mar 31, 2016
Content may be subject to copyright.
Ultrasonic Imaging
2014, Vol 36(1) 18 –34
© The Author(s) 2013
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0161734613508933
ultrasonicimaging.sagepub.com
Article
The Effect of Dead Elements on
the Accuracy of Doppler
Ultrasound Measurements
Jaromir Vachutka1, Ladislav Dolezal2, Christian Kollmann3,
and Jakob Klein3
Abstract
The objective of this study is to investigate the effect of multiple dead elements in an ultrasound
probe on the accuracy of Doppler ultrasound measurements. For this work, we used a specially
designed ultrasound imaging system, the Ultrasonix Sonix RP, that provides the user with the ability
to disable selected elements in the probe. Using fully functional convex, linear, and phased array
probes, we established a performance baseline by measuring the parameters of a laminar parabolic
flow profile. These same parameters were then measured using probes with 1 to 10 disabled
elements. The acquired velocity spectra from the functional probes and the probes with disabled
elements were then analyzed to determine the overall Doppler power, maximum flow velocity,
and average flow velocity. Color Flow Doppler images were also evaluated in a similar manner. The
analysis of the Doppler spectra indicates that the overall Doppler power as well as the detected
maximum and average velocities decrease with the increasing number of disabled elements. With
multiple disabled elements, decreases in the detected maximum and average velocities greater than
20% were recorded. Similar results were also observed with Color Flow Doppler measurements.
Our results confirmed that the degradation of the ultrasound probe through the loss of viable
elements will negatively affect the quality of the Doppler-derived diagnostic information. We
conclude that the results of Doppler measurements cannot be considered accurate or reliable if
there are four or more contiguous dead elements in any given probe.
Keywords
Doppler measurements, flow test object, probe failure, dead elements, quality assurance
Introduction
Doppler ultrasound techniques are widely used in medical diagnostics of cardiovascular diseases.
The use of ultrasound requires both its safe application as well as a clinically efficacious diagno-
sis. Accurate spectral Doppler measurements are often of crucial quantitative diagnostic
importance, for example, when the peak systolic blood velocity is measured in patients with
1Department of Medical Biophysics, Faculty of Medicine and Dentistry, Palacky University, Olomouc, Czech Republic
2Medical Ultrasound Research Laboratory, Palacky University, Olomouc, Czech Republic
3Center for Medical Physics and Biomedical Engineering, Medical University of Vienna, Austria
Corresponding Author:
Jaromir Vachutka, Faculty of Medicine and Dentistry, Department of Medical Biophysics, Palacky University,
Hnevotinska 3, 775 15 Olomouc, Czech Republic.
Email: jaromir.vachutka@upol.cz
508933UIX36110.1177/0161734613508933Ultrasonic ImagingVachutka et al.
research-article2013
by guest on March 15, 2016uix.sagepub.comDownloaded from
Vachutka et al. 19
suspected arterial disease and that velocity then converted to a percentage stenosis using standard
tables.1,2 The quality of the diagnostic conclusions may depend on several factors such as the skill
of the operator of the ultrasound system or the patient’s body habitus. However, the quality or,
more importantly, the lack of quality of both the B-mode images and Doppler spectra may also
lead to either a misdiagnosis or underdiagnosis. It is not possible to obtain clinically efficacious
diagnostic information with an ultrasound imaging and Doppler system that does not perform
properly. Therefore, it is necessary to periodically objectively test the performance of the ultra-
sound system and its attendant probes. The purpose of evidence-based quality assurance (EBQA)
testing is to adequately characterize the specific performance parameters for imaging equipment.
The results are then compared to vendor specifications, published values, or previous measure-
ments to determine if the performance is adequate.3 Nevertheless, in the Czech Republic and in
Austria, there are no official EBQA programs for testing diagnostic ultrasound systems in this
manner. We are currently in the process of preparing a draft proposal of such an EBQA system
designed to address this deficit.
An ultrasound probe is one of the most important parts of the ultrasound imaging system. In
addition, because it is held by the operator during the examination, the incidence of defective
ultrasound transducers is relatively high. A Swedish study tested the performance of probes in
use in routine clinical practice using the Sonora First Call Test System.4 This particular study
revealed that 39.8% of 676 evaluated transducers were defective in some manner. The observed
transducers failures included delaminations (66.5%), breaks in the cable (21.2%), short circuits
(8.5%), and weak or dead elements (3.8%). A similar Finnish study tested 151 transducers with
17% found to be defective.5 The authors of the Finnish study used two additional methods to
assess the quality parameters: tissue-mimicking phantom measurements and simple visual checks
to identify any readily apparent physical damage. The authors concluded that using all three
methods produce partly complementary results and should be included in QA testing of ultra-
sound scanners and probes.
A number of phantoms have been developed over the years for Doppler ultrasound, for exam-
ple, flow phantoms (including anatomical phantoms and simulation of arterial disease), perfusion
phantom, string and belt phantoms, electronic signal injection phantoms, and vibrating target
phantom.6 Examples of QA programs for B-mode3,5,7-12 and Doppler techniques9,11-13 can be
found in the literature.
So there are multiple tools on the market designed, in one manner or another, to detect defec-
tive probes. But it is necessary to quantify how a given probe failure will affect the quality of the
resultant diagnostic information. The potential of element degradation to affect the clinical results
was qualitatively investigated by Weigang and Moore.14 They concluded that the overall health
of the transducer array is critical to obtaining a high-quality and efficacious ultrasound study.
However, the authors of the Finnish study point out that the number of weak or dead elements
required to have significant impact on the quality of diagnostic ultrasound with different trans-
ducer types should be studied more closely.5 No study to date has determined how many dead
elements in the probe can be accepted. There are currently no quantitative studies in the literature
focused on these issues despite the fact that defective transducers in clinical use are common.
The objective of our work is to quantitatively describe the effect of dead elements on the
accuracy of Doppler ultrasound measurements.
Materials and Methods
Simulation of Probe Failures
A specially designed ultrasound imaging system, Sonix RP (Ultrasonix Medical Corporation,
Vancouver British Columbia), was used in this study. This ultrasonographic device is equipped with
by guest on March 15, 2016uix.sagepub.comDownloaded from
20 Ultrasonic Imaging 36(1)
a linear array probe (L14-5/38: 128 elements; center frequency 7.2 MHz), a convex array probe
(C5-2/60: 128 elements; center frequency 3.5 MHz; radius 60 mm), and a phased array probe (PA4-
2/20: 64 elements; center frequency 2.5 MHz). The software of this device allows the user to elec-
tronically disable the elements in the probe. The disabled elements are functionally equivalent to
dead elements, that is, they can neither transmit nor receive ultrasound waves. By selectively dis-
abling elements, we can create various degrees of precisely defined probe failures. For the measure-
ments used in this study, we selected probe element failures that corresponded to the worst-case
acoustic conditions. We used groups of 1 to 10 consecutive dead elements located toward the mid-
dle of the aperture that are normally active during the spectral Doppler mode. The various elements
used and their respective positions in the probe aperture are illustrated in Figure 1.
The fully functional probes were tested using the Sonora First Call Test System.4 The sensitiv-
ity of piezoelectric elements in fully functional probes is shown in Figure 2.
Using both the fully functional and defective probes, we measured and recorded the parame-
ters of a laminar parabolic flow profile.
Doppler Flow Phantom
The parabolic flow was generated by a flow test device, which satisfies the requirements of the
IEC 61685 standard.15 The flow Doppler test device is composed of agar-based tissue-mimicking
material (TMM), blood-mimicking fluid (BMF), tubing (two tubes of inner diameter 4.0 and 8.0
Figure 1. Simulation of the various probe failure conditions—the active aperture used during the
spectral Doppler mode is highlighted in green, the dead elements are highlighted in red.
by guest on March 15, 2016uix.sagepub.comDownloaded from
Vachutka et al. 21
mm are embedded in the TMM and carry the BMF), and a flow system generating device (peri-
staltic pump Masterflex L/S 7550-30, pulse dampener Masterflex L/S 07596-20).16,17 The BMF
was prepared according to the guidance found in the work of Ramnarine et al.18 This type of BMF
was found to be suitable for the use in test devices designed for Doppler performance assess-
ment.19 The composition of this BMF is (in % weight) 5 µm polyamide particles—Orgasol
(1.82%), pure water (83.86%), pure glycerol (10.06%), dextran (3.36%), surfactant—Synperonic
N (0.9%), and antifungal agent.
Figure 2. Sensitivity of piezoelectric elements in fully functional probes measured using the Sonora First
Call Test System.
by guest on March 15, 2016uix.sagepub.comDownloaded from
22 Ultrasonic Imaging 36(1)
Measurement of Doppler Spectra
We used convex, linear, and phased array probes to investigate the flow parameters in the 8.0 mm
tube. The volumetric flow rate was set to 200 ml/min during all measurements. By using continu-
ous laminar parabolic flow profile (which is satisfied via the phantom construction), we know the
average velocity, νA, of the flow corresponds to 6.63 cm/s (νA = volumetric flow rate/cross-sec-
tional area of the tube) and the maximum velocity, νMAX, reaches 13.26 cm/s (νMAX = 2*νA). The
4.0 mm tube was not used in this study.
The probe was situated on the surface of the phantom in such a way that the longitudinal axis
of the tube was located in the azimuthal scanning plane. The Doppler spectra were measured at
the maximum possible depth beneath the phantom surface. This depth corresponded to 19 mm in
the case of linear array probe and to approximately 40 mm in the case of both the convex and
phased array probes (Figure 3).
The parameters of the imaging system were optimized to obtain the best possible Doppler
spectrum using a fully functional probe. We used the lowest possible transmit frequency (2 MHz
for the convex and phased array probes and 4 MHz for the linear array probe) to minimize the
attenuation of transmitted pulses and received Doppler signals. The pulse repetition frequency
(PRF) was set to the lowest value that provided a nondistorted record of the complete Doppler
spectrum (1.3 kHz for all probe types). The gate (sample volume length) was adjusted to cover
the entire cross-section of the tube. The Doppler transmit power was set to the maximum level
allowable. The Doppler gain was set to obtain the maximum Doppler signal without any visible
white noise (4% of the maximum possible gain for the linear array probe, 6% of the maximum
possible gain for the convex array probe, and 8% of the maximum possible gain for the phased
array probe). The wall filter was set to the lowest possible frequency to minimize the distortion
of the Doppler spectrum (24 Hz for convex and phased array probe and 48 Hz for the linear array
probe). The Doppler angle was set at 60°. The imaging focal zone was adjusted to the position of
the Doppler sample volume.
Figure 3. Experimental setup.
by guest on March 15, 2016uix.sagepub.comDownloaded from
Vachutka et al. 23
We collected and stored 10 images of the Doppler spectrum. The identical measurements were
repeated with 1 to 10 dead elements for each probe (the experimental setup, the position of the probe
on the surface of the flow test device, and the parameters of the imaging system remained the same).
Analysis of Doppler Spectra
The acquired Doppler spectra were analyzed. The image of each spectrum was converted to a
bitmap. From the pixel values, we selected an area of interest that covered the entire range of
measured velocities with a time period of 3 s. Within the area of interest, the pixel values corre-
sponding to the same value of measured velocity (these lie on horizontal lines) were averaged.
Using this approach, we obtained one histogram of measured velocities from each stored image.
The calculation of the histogram of measured velocities is based on the assumption that the flow
in the tube is continuous. Figure 4 illustrates the analysis of Doppler spectra.
We then calculated the average histogram of measured velocities from 10 images stored for a
particular probe failure and experimental setup. The 68.27% confidence interval was calculated.
The results obtained using the same experimental setup with different degrees of probe failures
were compared in the graph. We then calculated the overall Doppler power, the maximum veloc-
ity, and the average velocity from the average histogram of measured velocities. The calculations
were performed using MS Office Excel. Any changes of these parameters higher than 10% from
the fully functional probe results were considered significant. Results acquired for different probe
failures were also compared in the graph. Figure 5 illustrates the calculation of investigated param-
eters. As one can see from Figure 5 (gray line), negative velocities can be found in the average
histogram of measured velocities. We assume that these negative velocities can be considered as
an artifact arising during spectral Doppler measurements. However, these negative velocities were
included in the calculations of the overall Doppler power and the average velocity.
The pixel values in the Doppler spectrum are proportional to the Doppler power. Although it
is not possible to calculate the absolute value of the Doppler power in Watts from the spectrum,
the relative changes of this parameter can be evaluated. We identified an interval of velocities
with the average pixel value higher than 1/10 of the maximum from the average histogram of
measured velocities (blue line in Figure 5). The overall Doppler power was calculated as a sum
of all the average pixel values within this interval. The 68.27% confidence interval was then
calculated. The maximum detected velocity, νMAX, was determined as the maximum value of
velocity with the average pixel value higher than 10 (red line in Figure 5). The average velocity,
vA, was calculated as a weighted average of velocities with the average pixel value higher than
1/10 of the maximum (blue line in Figure 5). The average pixel values are the weights for the
corresponding values of the velocity. The 68.27% confidence interval was then calculated.
Color Flow Doppler
Similarly to the changes in the spectral Doppler measurements, the Color Flow Doppler modali-
ties are also negatively influenced by defective transducer elements. To investigate these changes,
we included a small set of Color Flow Doppler measurements with different probe types and
measurement settings. Each set of measurements was done using a fully functional probe and
then using a probe with four and eight simulated dead elements without relocating the probe and
changing the parameters of the ultrasound system. The screenshots of the Color Flow Doppler
data were converted from the RGB into the HSV color representation model (Image Magick
Vers. 6.5.7-8). Further, only the hue channel was analyzed as a good approximation of the hue-
encoded velocity information. For the signal amplitude determination, all pixels containing color
information were counted, and for the spectral analysis, the 1-byte-normalized hue information
by guest on March 15, 2016uix.sagepub.comDownloaded from
24 Ultrasonic Imaging 36(1)
was used directly. For the described Doppler data processing steps, we used our developed in-
house program running under LINUX. We evaluated three different parameters: the total signal
amplitude (sensitivity) as the number of display pixels, the highest gray value in the hue channel
(corresponding to the maximum measured velocity), and the average gray value in the hue chan-
nel (corresponding to the average measured velocity).
Results
Doppler Spectrum
The effect of dead elements on the Doppler spectra is demonstrated by the results obtained using
the phased array probe PA4-2/20.
Figure 4. Analysis of Doppler spectra.
by guest on March 15, 2016uix.sagepub.comDownloaded from
Vachutka et al. 25
Figure 6 shows average histograms of measured velocities. One can see that the level of the
acquired signal gradually decreases with an increasing number of dead elements. The maximum
of the histogram moves to the lower velocities and the histogram itself becomes narrower. This
change of the histogram shape also affects other investigated parameters (maximum and average
detected velocity).
Figure 7 describes the change of the overall relative Doppler power. The overall relative Doppler
power decreases gradually with an increasing number of dead elements. We can see that the
decrease of the Doppler power is higher than 10% when the number of contiguous dead elements
0
20
40
60
80
100
120
140
-20-10 0102030
PIXEL VALUE
[1/256]
v [cm/s]
0
1
2
3
4
5
6
7
8
9
10
NUMBER OF
DEAD ELEMENTS
Figure 6. Doppler spectrum—average histograms of measured velocities (PA4-2/20; f = 2 MHz).
Figure 5. Calculation of investigated parameters.
by guest on March 15, 2016uix.sagepub.comDownloaded from
26 Ultrasonic Imaging 36(1)
is equal to or greater than two. The relative Doppler power detected by the defective probe with 10
dead elements is only one-third of the Doppler power detected by the fully functional probe.
The change of the maximum detected velocity due to the increasing severity of the probe
failure is shown in Figure 8. The decrease of the maximum velocity higher than 10% was seen
using the probe with 8 dead elements, the decrease by 15% was seen using the probe with 10 dead
elements.
Figure 9 shows the change of the average velocity with an increasing number of dead ele-
ments. The decrease of the average velocity by 12% was found for the probe with 10 dead ele-
ments. This decrease of the average velocity corresponds to the change of the shape of the
velocity histogram.
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
024681012
RELATIVE DOPPLER POWER
DEAD ELEMENTS
DECREASE BY 10 %
Figure 7. Change of the overall relative Doppler power with an increasing number of dead elements
(PA4-2/20; f = 2 MHz).
14
15
16
17
18
024681012
vMAX
[cm/s]
DEAD ELEMENTS
DECREASE BY 10 %
Figure 8. Change of the maximum detected velocity with an increasing number of dead elements
(PA4-2/20; f = 2 MHz).
by guest on March 15, 2016uix.sagepub.comDownloaded from
Vachutka et al. 27
Comparison of Probes
The Doppler spectra obtained using the fully functional convex, linear, and phased array probes
are compared in Figure 10.
Figures 11, 12, and 13 compare the change of the overall relative Doppler power, the maxi-
mum detected velocity, and the average velocity, respectively, due to an increasing number of
dead elements for convex (C), linear (L), and phased array (PA) probes. The horizontal lines
correspond to the change of investigated parameters by 10%. Table 1 compares the number of
dead elements that causes the change of investigated parameters higher than 10% when using
different types of probes. Table 2 summarizes the change of investigated parameters in the case
of 10 dead elements.
7.0
7.2
7.4
7.6
7.8
8.0
8.2
8.4
8.6
8.8
9.0
024681012
vA
[cm/s]
DEAD ELEMENTS
DECREASE BY 10 %
Figure 9. Change of the average velocity with an increasing number of dead elements (PA4-2/20;
f = 2 MHz).
0
20
40
60
80
100
120
140
160
180
-20-10 0102030
PIXEL VALUE
[1/256]
v [cm/s]
CONVEX
LINEAR
PHASED ARRAY
Figure 10. Comparison of probes—Doppler spectra (fully functional probes).
by guest on March 15, 2016uix.sagepub.comDownloaded from
28 Ultrasonic Imaging 36(1)
0
1000
2000
3000
4000
5000
6000
7000
024681012
RELATIVE DOPPLER POWER
DEAD ELEMENTS
C
L
PA
Figure 11. Comparison of probes—change of the overall relative Doppler power.
12
14
16
18
0246810 12
vMAX
[cm/s]
DEAD ELEMENTS
C
L
PA
Figure 12. Comparison of probes—change of the maximum detected velocity.
5
6
7
8
9
024681012
vA
[cm/s]
DEAD ELEMENTS
C
L
PA
Figure 13. Comparison of probes—change of the average velocity.
by guest on March 15, 2016uix.sagepub.comDownloaded from
Vachutka et al. 29
We can see that the decrease in the overall relative Doppler power with an increasing number
of dead elements is slightly higher for convex array probes than for linear and phased array
probes (Figure 11). The decrease in the Doppler power higher than 10% was caused by only one
dead element when using a convex array probe. The decrease of the maximum detected velocity
was highest for the convex array probe (Figure 12). A decrease higher than 10% was found in the
case of 4 dead elements, a decrease by 23% in the case of 10 dead elements. By contrast, the
effect of the probe failure on the value of the maximum detected velocity is much less significant
when using a linear array probe (maximal decrease by 6.5% was found). Similar results were also
determined in the case of the average velocity (Figure 13). Maximal influence of this parameter
was again found in the case of the convex array probe (a decrease higher than 10% in the case of
5 dead elements, maximal decrease by 20%) and only slight influence was found in the case of
the linear array probe (maximal decrease by 6%).
Color Flow Doppler
Figure 14 illustrates the changes in the Color Flow Doppler data due to the increasing degree of
the probe failure. The change of investigated parameters is summarized in Table 3.
The results in the table show that the signal amplitude obtained by a defective probe is reduced
roughly proportionally to the magnitude of the defect. Similarly, as in the case of the spectral
Doppler measurements, the maximum velocity as well as the average velocity generally decrease
with an increasing number of dead elements. A slight increase of these parameters was observed
when using linear and convex array probe with four dead elements. The effect of probe failures
on the Color Flow Doppler data was most significant in the case of the phased array probe.
Discussion
Doppler Spectrum
If we consider continuous laminar parabolic flow profile and complete and uniform insonation of
the whole tube cross-section, we should detect a constant signal level for all velocities between
zero and maximum velocity. However, the shape of the velocity histograms obtained using the
phased array probe is different (Figure 6). The relatively lower level of the signal detected for
Table 1. The Number of Dead Elements that Causes the Change of Investigated Parameters Higher
than 10% (Spectral Doppler Measurements).
Convex Probe Linear Probe Phased Array Probe
Doppler power 1 2 2
Maximum velocity 4 >10 8
Average velocity 5 >10 9
Table 2. The Change of Investigated Parameters in the Case of 10 Dead Elements (Spectral Doppler
Measurements).
Convex Probe (%) Linear Probe (%) Phased Array Probe (%)
Doppler power −74 −58 −66
Maximum velocity −23 −6.5 −15
Average velocity −20 −6 −12
by guest on March 15, 2016uix.sagepub.comDownloaded from
30 Ultrasonic Imaging 36(1)
lower velocities can be explained by nonuniform insonation of the tube cross-section. This phe-
nomenon was studied by Thompson et al. 20 and Steel and Fish.21,22
The value of maximum detected velocity measured by the fully functional phased array probe
(vMAX = 17.48 cm/s) is approximately 30% higher than the true maximum velocity in the tube of
the Doppler phantom (13.26 cm/s). This discrepancy can be explained by a phenomenon known
as intrinsic spectral broadening.23 This effect is associated with the size of the active aperture
used to generate and receive the Doppler beam and leads to the overestimation of the maximum
velocity derived from maximum Doppler frequency. Hoskins derived that the error, verr, in esti-
mated maximum velocity, which is defined as verr = (vestimated − vtrue) / vtrue, can be calculated
using the equation verr = (D/2L)·tan(θ).6 D denotes the length of the active aperture in a linear
array, L denotes the depth of the measurement, and θ is the Doppler angle. In our case, L = 40
mm, θ = 60°, and verr ≈ 0.3. The length of the corresponding aperture calculated using these val-
ues is approx. 14 mm. The distance between the centers of two consecutive elements (the pitch)
in the phased array probe is 0.3 mm and the calculated length of the aperture corresponds to 47
elements. The phased array probe usually uses all the elements in the array to form transmission
and receive beams (64 in our case). The overestimation of the maximum detected velocity can be
thus explained by this phenomenon. The maximum Doppler frequency shift is detected at one
extreme edge of the aperture, and the level of the received signal is relatively low. As was shown
in Figure 7, the backscattered Doppler power decreases with an increasing number of dead ele-
ments. This leads to the decrease of the level of the received signal which results in the loss of
the signal with the maximum Doppler shift. Therefore, the maximum detected velocity decreases
with an increasing number of dead elements (Figure 8).
Figure 14. Change of the Color Flow Doppler data (linear array probe).
Table 3. The Percent Change of Investigated Parameters (Color Doppler).
Dead
Elements
Convex Probe
(%)
Linear Probe
(%)
Phased Array
Probe (%)
Total signal amplitude 4 −3.7 +2.4 −3.5
8 −14.0 −9.1 −5.6
Maximum velocity 4 −3.5 +2.7 −5.0
8 −8.2 −0.0 −9.4
Average velocity 4 +4.5 +4.9 −1.6
8 −6.0 −3.4 −10.7
by guest on March 15, 2016uix.sagepub.comDownloaded from
Vachutka et al. 31
Similarly as in the case of the maximum detected velocity, the value of the average velocity
measured by the fully functional phased array probe (vA = 8.538 ± 0.008 cm/s) is overestimated
when compared with the true value in the tube of the Doppler phantom (6.63 cm/s). This is
caused primarily by the nonuniform insonation of the tube cross section (i.e., the flow signal is
not detected from the entire area near the tube wall where the fluid moves significantly slower
than in the center of the tube). The second reason for the overestimation of the average velocity
is that the ultrasound system’s wall filter removes the low-frequency signal from the Doppler
spectrum. However, this filter was set to the lowest possible cutoff frequency during all the mea-
surements in this study. The decrease in the average velocity due to the increasing degree of the
probe failure (Figure 9) is a consequence of the decrease in the maximum detected velocity.
Comparison of Probes
We discovered some differences between the average histograms of measured velocities obtained
using different types of probes despite the fact that the experimental setup remained the same
(Figure 10). If we compare the shape of the Doppler spectra, we can deduce that the linear array
probe emits the narrowest ultrasound beam (there is a significant predominance of high velocities
in the average histogram of measured velocities). By contrast, the widest ultrasound beam is
emitted by the phased array probe (the shape of the spectrum is closest to the theoretical predic-
tion). The width of the Doppler spectrum is not the same for all probe types. The differences
between the probes can be explained by the length of the active aperture. The calculated value of
the active aperture length is for the linear array probe (L = 19 mm, θ = 60°, vMAX = 16.01 cm/s,
verr ≈ 0.2, pitch = 0.3048 mm) approx. 4.4 mm and for the convex array probe (L = 38 mm, θ =
60°, vMAX = 15.80 cm/s, verr ≈ 0.2, pitch = 0.479 mm) approx. 8.8 mm. These active aperture
lengths correspond to 15 elements in the case of the linear array probe and to 19 elements in the
case of the convex array probe. All these results are consistent with usual active aperture lengths.
Furthermore, we determined that the width of the Doppler spectrum is dependent on the setting
of the overall Doppler gain level. The measurements with convex and phased array probes were
made at nearly the same depth beneath the phantom surface. The gain level was set to 6% and 8%
of the maximum possible gain during measurements with convex and phased array probes,
respectively. We verified that if the gain was set to 6% during the measurements with the phased
array probe, the width of the Doppler spectrum would be almost the same as in the case of the
measurement with the convex array probe.
The value of the average velocity measured by a fully functional convex array probe (vA =
7.270 ± 0.008 cm/s) is significantly lower than the value acquired by both linear and phased array
probes (Figure 13). This can be explained by the higher proportion of negative velocities as is
evident from the Doppler spectrum (Figure 10). We can observe that the average velocity
increases in the case of one and two dead elements when using a convex array probe. This result
is entirely opposite compared to all other findings. However, the increase in the average velocity
is very small and thus clinically irrelevant.
The probe failure influence on the Doppler ultrasound measurements accuracy is related to the
ratio of dead elements to functional elements in the active aperture. To obtain the aperture size,
we measured the ultrasound beam profile generated by each probe in the Doppler mode. The
analysis of these measurements enabled us to determine approximately the length of the transmit
aperture and the corresponding number of active elements. These results indicate that linear and
phased array probes use the same number of active elements (24) and convex array probe uses
only 20 active elements in the aperture. It was thus shown that the same number of dead elements
has higher effect in the case of smaller active aperture (the change of all investigated parameters
was most significant when using the convex array probe). Therefore, the severity of the probe
by guest on March 15, 2016uix.sagepub.comDownloaded from
32 Ultrasonic Imaging 36(1)
defects should be more precisely expressed as a ratio between the number of dead elements and
the number of all elements contained in the active aperture.
Conclusions
We can conclude that dead elements undoubtedly affect the accuracy of spectral Doppler mea-
surements. Our results revealed that in the vast majority of cases, all investigated parameters
(overall relative Doppler power, maximum detected velocity, and average velocity) decrease
with an increasing number of dead elements. The overall relative Doppler power is the most
affected parameter, conversely the average velocity is the least affected parameter. Our results
are consistent with conclusions of Weigang and Moore.14 But our results indicate that the effect
of dead elements on the accuracy of spectral Doppler measurements is not as significant as was
found by Weigang and Moore. They concluded that as few as two consecutive dead elements
in a 128-element array can negatively impact the overall quality and clinical efficacy of any
given examination. But we can conclude that the effect of three consecutive dead elements on
the accuracy of spectral Doppler measurements can be considered as negligible—the change
of maximum and average velocities higher than 10% was not observed during any measure-
ments when the number of dead elements was three or less. However, the decreased sensitivity
of the ultrasound system (which corresponds to the decrease of the overall Doppler power) can
negatively affect the results of clinical studies even in this case. If the number of dead elements
is four or higher, the results of Doppler measurements cannot, in general, be considered as
accurate.
We revealed significant differences between results obtained with different types of probes.
The convex array probe was found to be the most sensitive to the effects of probe failure on the
accuracy of spectral Doppler measurements. By contrast, the change of investigated parameters
was least significant when using the linear array probe. This implies that the extent of the influ-
encing of the accuracy of Doppler measurements could be related to the arrangement of the ele-
ments in the probe (linear versus curvilinear array). Furthermore, the results of Doppler
measurements were found to be dependent on the geometry of the ultrasound beams. The next
step of our research will thus be the measurement of the ultrasound field generated by fully func-
tional versus defective probes.
The evaluation of Color Flow Doppler data confirmed that this ultrasound mode is also con-
siderably affected by defective probe elements. Although the Color Flow Doppler modes are not
generally used for quantitative velocity measurements, the reduction of the total signal amplitude
can lead to important clinical consequences (e.g., loss of Color Flow Doppler data from small
vessels or the impossibility of detecting very slow flows).
Our results confirmed that it is necessary to periodically test the proper function of ultrasound
imaging systems because the degradation of the ultrasound probe negatively affects the quality
of the diagnostic information. We believe that our method of Doppler spectra analysis is appro-
priate for the evaluation of the degree of ultrasound probe failures and their potential impact on
the diagnosis. Our evaluation method demonstrates that the detected changes of the flow param-
eters are only caused by the effects of dead elements (the experimental setup including the posi-
tion of the probe on the surface of the flow phantom remains the same during the whole set of
measurements). We believe that this new method brings new possibilities in the approach to
evidence-based quality assurance testing of ultrasound systems. Our results showed that this
method is very sensitive to performance variances, and therefore, we are able to detect even
slight changes in the Doppler spectra. However, the application of this method is based on the
assumption that we are able to electronically create precisely defined probe failures, which is not
true for most ultrasound imaging systems commonly used in clinical practice. Furthermore, the
by guest on March 15, 2016uix.sagepub.comDownloaded from
Vachutka et al. 33
observed effects can be dependent on other parameters that were not evaluated in this study (e.g.,
on the size of the active aperture).
Our ultimate goal is to establish a quantitative threshold probe defect level which would deter-
mine that a given device should not be used in a clinical practice. Based on the results of this
particular study, we can preliminarily recommend that in the case of Doppler ultrasound mea-
surements, the threshold probe failure corresponds to four consecutive dead elements. This rec-
ommendation is valid for ultrasound scanners of similar technical level as we used for our
research. Nevertheless, this issue should continue to be investigated in more detail in subsequent
studies (e.g., we have to find how probe failures affect the measurements of pulsatile flow).
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or
publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publi-
cation of this article: This work was supported by a project LF_2013_006.
References
1. Hoskins PR. A review of the measurement of blood velocity and related quantities using Doppler ultra-
sound. Proc Inst Mech Eng. 1999;213:391-400.
2. Grant EG, Benson CB, Moneta GL, Alexandrov AV, Baker JD, Bluth EI, et al. Carotid artery stenosis:
gray-scale and Doppler US diagnosis—Society of Radiologists in Ultrasound Consensus Conference.
Radiol. 2003;229:340-6.
3. Kofler JM Jr. Quality assurance of ultrasound imagers: procedures, expectations, and philosophies.
Handout. In: 43rd Annual Meeting of the American Association of Physicists in Medicine, Salt Lake
City, UT, 22-26 July 2001. Retrieved from http://www.aapm.org/meetings/2001AM/pdf/7194-74902.
pdf
4. Mårtensson M, Olsson M, Segall B, Fraser AG, Winter R, Brodin LA. High incidence of defective
ultrasound transducers in use in routine clinical practice. Eur J Echocardiogr. 2009;10:389-94.
5. Sipilä O, Mannila V, Vartiainen E. Quality assurance in diagnostic ultrasound. Eur J Radiol.
2011;80:519-25.
6. Hoskins PR. Simulation and validation of arterial ultrasound imaging and blood flow. Ultrasound Med
Biol. 2008;34:693-717.
7. Thijssen JM, Weijers G, de Korte CL. Objective performance testing and quality assurance of medical
ultrasound equipment. Ultrasound Med Biol. 2007;33:460-71.
8. Kollmann C.Apparative quality assurance of ultrasound imaging equipment. AUStrian Test Kit™.
2005. Retrieved from http://www.zmpbmt.meduniwien.ac.at/fileadmin/zbmtp/uploads/ultrasound/sci-
ence/austr_testkit_basic_v1engl.pdf
9. Thijssen JM, van Wijk MC, Cuypers MHM. Performance testing of medical echo/Doppler equipment.
Eur J Ultrasound. 2002;15:151-64.
10. Gibson NM, Dudley NJ, Griffith K. A computerised quality control testing system for B-mode ultra-
sound. Ultrasound Med Biol. 2001;27:1697-711.
11. Dudley NJ, Griffith K, Houldsworth G, Holloway M, Dunn MA. A review of two alternative ultra-
sound quality assurance programmes. Eur J Ultrasound. 2001;12:233-45.
12. Dolezal L, Kollmann C. Patient—Ultrasound Interaction. Olomouc, Czech: Palacky University; 2010.
13. Sperrin MW, Mutch S. The assessment of a Doppler ultrasound phantom with an aim to establish a
routine quality assurance programme. Eur J Ultrasound. 1996;4:63-7.
14. Weigang B, Moore GW, Gessert J, Phillips WH, Schafer M. The methods and effects of transducer
degradation on image quality and the clinical efficacy of diagnostic sonography. J Diagn Med Sonog.
2003; 19:3-13.
by guest on March 15, 2016uix.sagepub.comDownloaded from
34 Ultrasonic Imaging 36(1)
15. International Electrotechnical Comission 61685. Ultrasonics—Flow Measurement Systems—Flow
Test Object. Geneva, Switzerland: International Electrotechnical Comission; 2001.
16. Teirlinck CJPM, Bezemer RA, Kollmann C, Lubbers J, Hoskins PR, Fish P, et al. Development of an
example flow test object and comparison of five of these test objects, constructed in various laborato-
ries. Ultrasonics. 1998;36:653-60.
17. Kollmann C, Bezemer RA, Fredfelt KE, Schaarschmidt UG, Teirlinck CJPM, et al. Ein Testobjekt
für die apparative Qualitätssicherung bei Ultraschall-Doppler(Duplex)-Geräten, ausgehend vom
Normenentwurf IEC 61685 [A Laminar Flow Test Object Based on the Draft IEC 61685 Standard for
Quality Assurance of Ultrasound Doppler Equipment]. Ultraschall in der Medizin. 1999;20:248-57.
18. Ramnarine KV, Nassiri DK, Hoskins PR, Lubbers J. Validation of a new blood-mimicking fluid for use
in Doppler flow test objects. Ultrasound Med Biol. 1998;24:451-9.
19. Ramnarine KV, Hoskins PR, Routh HF, Davidson F. Doppler backscatter properties of a blood-mim-
icking fluid for Doppler performance assessment. Ultrasound Med Biol. 1999;25:105-10.
20. Thompson RS, Bambi G, Steel R, Tortoli P. Intraluminal ultrasound intensity distribution and back-
scattered Doppler power. Ultrasound Med Biol. 2004;30:1485-94.
21. Steel R, Fish PJ. Lumen pressure within obliquely insonated absorbent solid cylindrical shells with
application to Doppler flow phantoms. IEEE Trans Ultrason Ferroelectr Freq Control. 2002;49:271-80.
22. Steel R, Fish PJ. A simulation of sample volume sensitivity for oblique pulsed finite beam insonation
of Doppler ultrasound flow phantom cylindrical vessels. IEEE Trans Ultrason Ferroelectr Freq Control.
2003;50:58-67.
23. Thrush AJ, Evans DH. Intrinsic spectral broadening: a potential cause of misdiagnosis of carotid artery
disease. J Vasc Investig. 1995;1:187-92.
by guest on March 15, 2016uix.sagepub.comDownloaded from
