Abstract

Aims

The objective was to follow-up the study ‘High incidence of defective ultrasound transducers in use in routine clinical practice’ and evaluate if annual testing is good enough to reduce the incidence of defective ultrasound transducers in routine clinical practice to an acceptable level.

Methods and results

A total of 299 transducers were tested in 13 clinics at five hospitals in the Stockholm area. Approximately 7000–15 000 ultrasound examinations are carried out at these clinics every year. The transducers tested in the study had been tested and classified as fully operational 1 year before and since then been in normal use in the routine clinical practice. The transducers were tested with the Sonora FirstCall Test System. There were 81 (27.1%) defective transducers found; giving a 95% confidence interval ranging from 22.1 to 32.1%. The most common transducer errors were ‘delamination’ of the ultrasound lens and ‘break in the cable’ which together constituted 82.7% of all transducer errors found. The highest error rate was found at the radiological clinics with a mean error rate of 36.0%. There was a significant difference in error rate between two observed ways the clinics handled the transducers. There was no significant difference in the error rates of the transducer brands or the transducers models.

Conclusion

Annual testing is not sufficient to reduce the incidence of defective ultrasound transducers in routine clinical practice to an acceptable level and it is strongly advisable to create a user routine that minimizes the handling of the transducers.

Introduction

Quality control within medical ultrasound has been addressed in the literature for many years, but is still a topic for further research.1–4 In an earlier study5 (henceforth called ‘the previous study’), we found that an unacceptable high incidence (39.8%) of defective ultrasound transducers were in use in routine clinical practice. Defective transducers were also considered to be the main cause of image degradation in an international survey.6 In the previous study, we showed that heart diseases can be missed due to a defective transducer which may lead to potential patient hazard. In order to meliorate the situation in the clinics, we advocated that more extensive testing of the ultrasound transducers is of utmost importance. It was however not possible from the previous study to state how often an ultrasound transducer should be tested in order to guarantee that it is in working order. Furthermore are there no general recommendations in this matter neither from the transducer manufacturers nor the manufacturers of testing systems. Quality control guidelines for ultrasound systems are given by several organizations, however, without specific transducer guidelines.7–9 Probably, one single recommendation valid for all transducers and all clinics is almost impossible because the active duty time of the ultrasound transducers can vary a lot between clinics and also between transducers of different applications. This problem results in that every clinic has to work out their own transducer quality assurance protocol based on their unique situation to meet their quality demands.

As a result of the previous study, many clinics realized the importance of transducer testing and the risk of missing diagnoses, and therefore introduced transducer testing to their quality assurance protocols. It was of course introduced without any knowledge about how frequently the tests had to be performed to assure the function of the transducers. A common choice among the clinics where we perform the tests was annual intervals, which also seems to be widely used internationally.6

In this follow-up study, we have compiled the results of having used a test interval of 1 year in 13 clinics, at five hospitals in the Stockholm area, with relatively high workload on their ultrasound transducers. The main goal of the study was to investigate if 1 year test interval is enough to decrease the number of defective ultrasound transducers to an acceptable level. Sub-goals of the study were to point out any significant differences in the rate of error between different transducer models, clinics, or clinic types.

Methods

Testing protocol

The transducers were tested with the Sonora FirstCall Test System. This is a test system where the transducer to be tested is connected to the test system instead of the ultrasound scanner. The function of each individual element in the transducer is tested one at a time. The test is performed in water, where the elements are activated and a metal target plate is used to reflect the ultrasound pulse emitted by each activated element. The returning pulse, received by the same element, is analysed by means of the peak-to-peak amplitude, centre frequency, pulse width, bandwidth, and the pulse waveform. The test system also measures the accumulated capacitance of every element and its wires to check for electrical failures.

The transducers

A total of 299 transducers were tested in this follow-up study. Every transducer at the clinics except single element and 3D transducers were tested. The transducers had 1 year before the study either passed the test during the annual test routine of that year or been put in to service as a replacement for a defective one. Since that test the transducers had been in normal use in the clinics. The sonographers using the transducers were unaware of the study going on.

The transducers originated from three different manufacturers: Siemens, GE, and Philips. The number of transducers from each manufacturer was 227 from Siemens, 59 from GE, and 13 from Philips. The numbers of transducer models from Siemens, GE, and Philips were 22, 15, and 4, respectively.

The clinics

The clinics in this study were also included in our previous study.5 They have since then introduced annual testing of their ultrasound transducers. The ultrasound transducers were tested at 13 clinics in five different hospitals. The number of ultrasound examinations performed at the clinics the year before the tests were performed varied from 6900 to 15 000 with a mean and standard deviation of 10 400 and 2500, respectively. The specialties of the clinics were radiology, cardiology, clinical physiology, obstetrics, and gynaecology.

Acceptance criteria and transducer errors

In this study, we used the same acceptance criteria as in our previous study. Thus, the following definitions of functionally acceptable element, weak element, and dead element were used: The sensitivity of an element is a measurement of the signal amplitude of the elements within the transducer. The amplitude value is related to the selected gain setting, so there is no definitive threshold for the sensitivity value. The sensitivity reflects both the ability of the elements to transmit and receive and the performance of the matching layer/ultrasound lens.

  • Functionally acceptable element: element with a sensitivity value of over 75% of the mean value for all elements within a transducer.

  • Weak element: element with a sensitivity value of between 10 and 75% of the mean value for all elements within a transducer.

  • Dead element: element with a sensitivity value of below 10% of the highest value within a transducer.

The transducers were considered defective if they contained more than four contiguous weak elements, three or more dead elements, or two contiguous dead elements.

The transducer errors were classified as: delamination, break in the cable, short circuit, weak, or dead elements. Delamination occurs when the backing material, the matching layer or the lens detaches from one or more elements and air or liquid has penetrated the transducer layers.

Data presentation and statistical analysis

The transducer errors are presented as percentage, absolute numbers and with a 95% confidence interval. Fisher's exact tests were performed to determine whether the prevalence of transducer errors differed significantly, both between transducers from the three different manufacturers, transducer models, clinics, and clinic types. To statistically compare the different types of clinics, the clinics were grouped into three categories; radiology, cardiology/clinical physiology, and obstetrics/gynaecology. The number of clinics in each group was as follows: four radiology, six cardiology/clinical physiology, and three obstetrics/gynaecology. The clinics, were also grouped into two categories, by the way they handled their transducers. The clinics in handling category one had their transducers constantly plugged into the ultrasound scanners and the transducers were hanging on a suspension device on the scanners when not in use. The clinics in handling category two had their transducers disconnected from the scanners when not in use. Between the examinations the transducers are transferred to, and stored in, a special transducer storage place at the clinic.

Results

Transducer errors

Of the 299 tested transducers, 81 were defective; giving an error rate of 27.1% with a 95% confidence interval ranging from 22.1 to 32.1%. This is significantly lower (P < 0.001) compared with the findings in the previous study5 where the 95% confidence interval for the error rate ranged from 36.1 to 43.5%.

There was a significant difference in the error rate of the five error types (P < 0.001). The distribution of the error types was as follows: 40 (13.4%) delamination, 27 (9.0%) break in the cable, 12 (4.0%) weak elements, 1 (0.3%) short circuit, and 1 (0.3%) dead elements.

Comparison of the manufacturers

Of the 227 transducers from Siemens, 67 (29.5%) were defective; 31 with delamination problems, 23 with broken signal wires, 12 with weak elements, and 1 with short-circuited signal wires. The 95% confidence interval of the error rate for the Siemens transducers ranged from 23.6 to 35.4%. Of the 59 transducers from GE, 11 (18.6%) were defective; 8 with delamination and 2 with broken wires and 1 with dead element. The 95% confidence interval of the error rate for the GE transducers ranged from 8.7 to 28.6%. Of the 13 transducers from Philips, 3 (23.1%) were defective; 2 with broken wires and 1 delaminated. The 95% confidence interval of the error rate for the Philips transducers ranged from 0.0 to 46.0%. There was no significant difference in the error rates of the three manufacturers (P > 0.05).

Comparison of the transducer models

The number of transducers of a specific model varied from 1 to 35 and the error rate from 0 to 100%. There was no significant difference in the error rates of the 41 transducer models included in the study (P > 0.05). Nor was it any significant differences when calculated among the transducers from each manufacturer. Table 1 shows the transducer model, number of each model and the percentage defective of each model.

Table 1

The transducer models, number and percentage defective transducers for each model

Transducer model Number Percentage defective 
Siemens 
 4V1 34 47.1 
 4V1c 66.7 
 4V2 13 15.4 
 15L8 18 11.1 
 8V5 
 8L5 35 28.6 
 4C1 33 48.5 
 6C2 27 33.3 
 3V2c 18 22.2 
 6L3 17 29.4 
 V5M 
 EV8C4 
 8C4 33.3 
 7V3 
 L7 
 V4 
 L5 
 L10 
 A7 
 C3 
 5V2c 
 9L4 
GE 
 M3S 29 17.2 
 M12L 33.3 
 7Lc 
 6-Tb 100 
 5S 
 3S 
 348C 100 
 10LB 
 3S-RS 
 4C-RS 100 
 5C 
 RAB4-8P 33.3 
 AB2-7 33.3 
 9L 
 M4S 
Philips 
 S5-1 
 C5-2 33.3 
 C8-4V 50 
 S8-3 
Transducer model Number Percentage defective 
Siemens 
 4V1 34 47.1 
 4V1c 66.7 
 4V2 13 15.4 
 15L8 18 11.1 
 8V5 
 8L5 35 28.6 
 4C1 33 48.5 
 6C2 27 33.3 
 3V2c 18 22.2 
 6L3 17 29.4 
 V5M 
 EV8C4 
 8C4 33.3 
 7V3 
 L7 
 V4 
 L5 
 L10 
 A7 
 C3 
 5V2c 
 9L4 
GE 
 M3S 29 17.2 
 M12L 33.3 
 7Lc 
 6-Tb 100 
 5S 
 3S 
 348C 100 
 10LB 
 3S-RS 
 4C-RS 100 
 5C 
 RAB4-8P 33.3 
 AB2-7 33.3 
 9L 
 M4S 
Philips 
 S5-1 
 C5-2 33.3 
 C8-4V 50 
 S8-3 

The transducers were considered defective if they contained more than four contiguous weak elements, three or more dead elements, or two contiguous dead elements (see section Acceptance criteria and transducer errors).

Comparison of the clinics and clinic types

The error rates in the clinics were ranging from 0 to 51.5%, but the differences between the 13 clinics were not significant (P > 0.05). It was however significant when the 13 clinics were divided into the three clinic types (P = 0.015). The highest error rate was found at the radiological clinics with a mean error rate of 36.0%. The mean error rate at the clinics of cardiology/physiology and obstetrics/gynaecology was 20.5 and 31.8%, respectively.

When comparing the two handling categories, the evidence for a statistical difference was even stronger (P = 0.002). The clinics belonging to category 2 had an error rate of 37.9% whereas category 1 had 21.4%. The 95% confidence intervals for category 1 and 2 ranged from 14.3 to 25.5% and 31.2 to 50.3%, respectively. Table 2 shows the transducer distribution in the clinics and the error rate of each clinic.

Table 2

The transducer distribution in the clinics and the error rate of each clinic

Clinic Number of ultrasound examinations 2008–2009 Number of transducers Handling category Percentage defective at the clinic 
Radiology 1 9100 32 31.3 
Radiology 2 8900 12.5 
Radiology 3 14 000 38 31.6 
Radiology 4 15 000 33 51.5 
Cardiology and clinical physiology 1 6900 25 
Cardiology and clinical physiology 2 10 400 28.6 
Cardiology and clinical physiology 3 7500 32 28.1 
Cardiology and clinical physiology 4 9400 36 22.2 
Cardiology and clinical physiology 5 8600 38 21.1 
Cardiology and clinical physiology 6 8600 28 25.0 
Obstetrics and gynaecology 1 11 500 28.6 
Obstetrics and gynaecology 2 12 500 37.5 
Obstetrics and gynaecology 3 12 500 28.6 
Clinic Number of ultrasound examinations 2008–2009 Number of transducers Handling category Percentage defective at the clinic 
Radiology 1 9100 32 31.3 
Radiology 2 8900 12.5 
Radiology 3 14 000 38 31.6 
Radiology 4 15 000 33 51.5 
Cardiology and clinical physiology 1 6900 25 
Cardiology and clinical physiology 2 10 400 28.6 
Cardiology and clinical physiology 3 7500 32 28.1 
Cardiology and clinical physiology 4 9400 36 22.2 
Cardiology and clinical physiology 5 8600 38 21.1 
Cardiology and clinical physiology 6 8600 28 25.0 
Obstetrics and gynaecology 1 11 500 28.6 
Obstetrics and gynaecology 2 12 500 37.5 
Obstetrics and gynaecology 3 12 500 28.6 

Discussion

In quality control of ultrasound scanners, two important decisions have to be made, first, what actions that has to be done, secondly, how often these actions need to be performed. Transducer testing is undoubtedly a necessary action in the quality assurance protocols of ultrasound scanners.5 How often it should be performed in order to guarantee the transducer function has not yet been fully elucidated. In the present study, the results of introducing annual transducer testing in clinics performing about 10 000 ultrasound examinations every year were presented. In addition to these annual tests the manufacturers conducted their normal regular maintenance to the ultrasound scanners in accordance with earlier existing service agreements; the transducer testing was considered by the clinics as a complement to these service agreements. The clinics were part of the previous study and these tests are the third annual tests since then. All the defective transducers found during this period have been replaced by new ones. It must be pointed out that the findings and the conclusions of this study are strongly linked to the workload of the ultrasound transducers included in this study; the situation might be very different in clinics where more or fewer examinations are performed.

In the previous study,5 where no regular transducer testing was performed by the concerned clinics, we found that the 95% confidence interval of the number of defective ultrasound transducers ranged from 36.1 to 43.5%. In this study, 3 years after the introduction of annual transducer testing, the 95% confidence interval was lower and ranged from 22.1 to 32.1%. That difference is by statistical definition significantly lower; the introduction of annual transducer testing has reduced the number of defective transducers in these clinics. However, the fact that the decrease is statistical significant does not mean much in this context; the error rate found is still unacceptable high. Having somewhere between one-fifth and one-third of the ultrasound transducers with unreliable function is an untenable situation that requires much more frequent testing of the transducers. The distribution of the error types was similar to the previous study with delamination and break in the cable as the most common errors; in this study these two errors together constituting 82.7% of the total errors compared with 87.7% in the previous study.

There was a significant difference between the three clinic types; the clinics from the cardiology/clinical physiology group had a significantly lower error rate than the clinics of radiology and obstetrics/gynaecology. There are different transducer models used at the different clinic types. For example, a majority of the transducers at the cardiology/clinical physiology clinics are phased array cardiac transducers, and curvilinear and vaginal transducers at the obstetrics/gynaecology clinics, although there are many more transducer types used at the radiological clinics. However, it is not possible to argue that the transducers would be the cause for the difference between the clinics since there is no significant difference in error rate between the transducer models. The radiological clinics were having the highest error rate. There was however one radiological clinic with much better error rate than the others, albeit based on fewer transducers. They were also different in another way; they handled their transducers differently than the other radiological clinics. Here, we would like to sound a warning note about the way some clinics are handling their transducers. During the tests, we have noticed two fundamentally different ways in the handling of the transducers. The first way, which was also the most common, is when the transducers constantly are plugged into the ultrasound scanners and are hanging on a suspension device on the scanners when not in use. The second way, used only at three radiological clinics, is a method where the transducers are disconnected from the scanners when not in use. Between the examinations the transducers are transferred to and stored in a special storage place at the clinic, which means that the transducers has to be transferred twice for every examination. This method leads inevitably to much more handling of the transducers and a subsequent higher risk of damaging the transducers. The method was used at the radiological clinics with the three highest error rates. The mean error rate at these clinics was 77.1% higher than the mean of all the other clinics in the study. Therefore, the high error rate at these radiological clinics probably not just depends on inadequate transducer testing, but also because of the way the transducers are handled between the examinations at these clinics.

In our studies, we have seen that it is difficult for the user to realize when the transducer function is deteriorating. We believe that it becomes even more difficult to realize it when the transducers are moved around to different scanners; the handling system does not provide the sonographers with a day-to-day experience of how the unit (ultrasound scanner and transducer) performs. Nor is it possible to predict when a transducer should be replaced. The defective transducers found are a combination of newer and older ones. Some need to be replaced after only 1 year of use whereas others are more than 5 years old and works perfectly normal. What determines the lifespan of a transducer is mainly how it is used, not how long it will be used.

When calculated for all 13 clinics, the differences in error rates were significant. There is particularly one clinic that stands out from the other 12. The best result found, was at a clinic in the clinical physiology group where no single defective transducer was found. This is undoubtedly a remarkable result and the underlying causes ought to be conveyed to other clinics so that they can improve their working methods accordingly. Such a considerable difference should reasonably be explained by the specific situation at that clinic. But we have not been able to find any parameter where this clinic is working differently from the other clinics; i.e. from the other clinics in category 1. Then one can assume that factors such as the training of the staff and the work load are of importance here. It is indeed the clinic in this study doing the least number of examinations every year. However, this difference in itself is too small to explain the better outcome for this clinic. Whatever the reasons are, this shows that an acceptable level can be achieved when the circumstances are favourable.

Furthermore, it proved to be an uneven distribution of the transducers both between manufacturers and models as shown in Table 1 and 2; there are many instances of these tables with small numbers. When the statistical calculations, performed in order to show if there are any statistically significant differences in error rates, are based on several instances with very small numbers, the calculations are more likely to give non-significant results. At first sight this might look like the cause for the non-significant results. But if the instances with very small numbers are excluded or combined with other groups, or in to new groups, the non-significant results remains. It is therefore not possible to comment if any transducer model or transducer from a specific manufacturer should be considered to be in greater need of testing; the correct conclusion about the transducers has to be that every transducer from every manufacturer is in great need of frequent testing.

The error rates are still unacceptable high with delamination of the ultrasound lens and break in the cable as the two most common error types. These are two error types that probably increase by rough handling since they are related to sensitive and relatively unprotected parts, such as the ultrasound lens and the thin signal wires within the cable. Furthermore, clinics with less extensive handling of their transducers are showing significantly lower error rates. We therefore, not just conclude that the transducers must be tested more often than once a year to guarantee fully functional transducers, but also strongly advice/recommend a handling system that minimizes the handling of the transducers.

It is probably appropriate to make an additional comment about the acceptance criteria that were used in the study. As in the previous study the transducers were considered defective if they contained more than four contiguous weak elements, two contiguous dead elements or more than two dead elements. Since then it has been pointed out that these criteria could be too restrictive. We fully agree in that statement; these acceptance criteria are by no means carved in stone and may need to be revised in the light of new knowledge. Today it is a matter of judgement where to put the limits of the criteria. It has been shown that two contiguous dead elements can result in a considerable increase in the magnitude of the side lobes in the beam profile and that six contiguous dead elements gives a pronounced loss in spectral Doppler sensitivity.10 This was shown for two specific transducer models; other transducer models may perhaps suffer less from dead elements through differences in the design, and might therefore need to be assessed under other premises. This uncertainty of where to put the limits of the criteria has not affected the results of the study in any significant way. There were very few of the defective transducers that were borderline cases. The two most common transducer errors, delamination and break in the cable, rarely affect only one or two elements but usually a much larger number of elements. Therefore, the conclusion that the transducers have to be tested more often than once a year to guarantee their function would remain the same even with a relatively large change of the acceptance criteria.

Funding

The study was funded by the Swedish Heart-Lung Foundation.

Conflict of interest: none declared.

References

1
Brendel
K
Filipczynski
LS
Gerstner
R
Hill
CR
Kossoff
G
Quentin
G
, et al.  . 
Methods of measuring the performance of ultrasonic pulse-echo diagnostic equipment
Ultrasound Med Biol
 , 
1977
, vol. 
2
 (pg. 
343
-
50
)
2
Hall
AJ
Fleming
JEE
A method of calibrating contact B-scanners
Ultrasonics
 , 
1978
, vol. 
16
 (pg. 
277
-
81
)
3
Gibson
NM
Dudley
NJ
Griffith
K
A computerised quality control testing system for B-mode ultrasound
Ultrasound Med Biol
 , 
2001
, vol. 
27
 (pg. 
1697
-
711
)
4
Thijssen
JM
Weijers
G
de Korte
CL
Objective performance testing and quality assurance of medical ultrasound equipment
Ultrasound Med Biol
 , 
2007
, vol. 
33
 (pg. 
460
-
71
)
5
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
, vol. 
10
 (pg. 
389
-
94
)
6
Shaw
A
Hekkenberg
R
NPL Report: Standards to Support Performance Evaluation for Diagnostic Ultrasound Imaging Equipment
 , 
2007
October 2007 ed.
Teddington
National Physical Laboratory
7
AIUM
Technical Standard Committee. Quality Assurance Manual for Gray-Scale Ultrasound Scanners, Stage 2
1995
AIUM
8
Price
R
IPSM Report No. 71 Routine quality assurance of ultrasound imaging systems
1995
The Institute of Physical Sciences in Medicin
9
Goodsitt
MM
Carson
PL
Witt
S
Hykes
DL
Kofler
JM
Jr
Real-time B-mode ultrasound quality control test procedures. Report of AAPM Ultrasound Task Group No. 1
Med Phys
 , 
1998
, vol. 
25
 (pg. 
1385
-
406
)
10
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 Sonography
 , 
2003
, vol. 
19
 (pg. 
3
-
13
)

Comments

0 Comments