Clinical implications of machineprobe ... - Wiley Online Library

43 downloads 13477 Views 82KB Size Report
Jul 13, 2012 - *Medical Physics and Clinical Engineering, Cambridge University Hospitals NHS Foundation Trust, Cambridge, UK; †Fetal Medicine, ... (CRL, FL, OFD) direction of the ultrasound beam, with the fetus ..... risk) devices and the process of self certification by ... Ultrasound Diagnostic Instrument, prosound α 7,.
Ultrasound Obstet Gynecol 2012; 40: 194–199 Published online 13 July 2012 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/uog.11081

Clinical implications of machine–probe combinations on obstetric ultrasound measurements used in pregnancy dating R. AXELL*, C. LYNCH†, T. CHUDLEIGH‡, L. BRADSHAW§, J. MANGAT¶, P. WHITE* and C. LEES† *Medical Physics and Clinical Engineering, Cambridge University Hospitals NHS Foundation Trust, Cambridge, UK; †Fetal Medicine, Cambridge University Hospitals NHS Foundation Trust, Cambridge, UK; ‡Rosie Ultrasound, Cambridge University Hospitals NHS Foundation Trust, Cambridge, UK; §Centre for Applied Medical Statistics, University of Cambridge, Cambridge, UK; ¶Medical Physics, University Hospitals of Leicester, Leicester, UK

K E Y W O R D S: accuracy; biparietal diameter; crown–rump length; femur length; fetal; growth; intraoperator; phantom; ultrasound equipment; variability

ABSTRACT Objective To investigate the effect of use of different machine–probe combinations on measurement variability and to assess how this variability impacts on accuracy of pregnancy dating. Methods Sixteen different ultrasound machine–probe combinations were used for lateral measurement of targets spaced 10 and 40 mm apart and axial measurement of targets spaced 10 and 50 mm apart in an ultrasound phantom, and differences between the measured and true values were determined. The mean of the 40-mm lateral measurements was used to estimate gestational age using standard obstetric dating tables for crown–rump length (CRL) and femur length (FL) and the mean of the 50-mm axial measurements was used to estimate gestational age using the obstetric dating tables for biparietal diameter (BPD). Results As absolute measurements became larger, differences observed using different machine–probe combinations exceeded those due to intraobserver variability. Maximum dating differences for first-trimester CRL of 2 days (based on a measured CRL range of 39–42 mm), second-trimester BPD of 4 days (based on a measured BPD range of 49–52 mm) and FL of 9 days (based on a measured FL range of 39–42 mm) were observed when measurements were transposed to the equivalent gestational age using standard charts. Conclusion Differences between measured and true values of biometric measurements reflect both machine– probe and intraobserver variability. Incremental firsttrimester CRL growth with time is rapid, but secondtrimester FL growth is much less so, leaving this lateral measurement more prone to both observer and machine–

probe errors. The only axial growth measurement commonly performed is BPD where the measurement differences were intermediate between those of CRL and FL. The differences that can be ascribed to different equipment combinations are in many cases greater than those expected in clinical practice and are of potential importance in determining how fetal biometry is used for dating pregnancies. Copyright  2012 ISUOG. Published by John Wiley & Sons, Ltd.

INTRODUCTION Fetal biometry measurements are performed as an integral part of most obstetric ultrasound examinations. In many instances, linear caliper ultrasound measurements are used to date the pregnancy. The measurement of choice for dating at 6–13 weeks’ gestation is the crown–rump length (CRL), with a reported accuracy of 3–4 days1 . After 13–25 weeks, femur length (FL), head circumference, occipitofrontal diameter (OFD) and biparietal diameter (BPD) measurements are normally used to estimate and assign gestational age2,3 . Accuracy of pregnancy dating is critical to subsequent obstetric decision-making in respect of, for example, assessing growth abnormalities or timing Cesarean section and induction of labor. Further, the accuracy of both the nuchal translucency and biochemistry components of Down syndrome risk assessment depends on the measured CRL being accurate. A difference of several days in the assigned gestational age of the fetus for a given nuchal translucency measurement can substantially alter the calculated risk for Down syndrome3 . Dating measurements are normally performed with the fetus orientated so that the measurement is taken in an axial (BPD) or lateral

Correspondence to: Dr C. Lees, Fetal Medicine, Rosie Maternity, Addenbrookes Hospital, Cambridge University Hospitals NHS Foundation Trust, Hills Road, Cambridge, CB2 2QQ, UK (e-mail: [email protected]) Accepted: 7 December 2011

Copyright  2012 ISUOG. Published by John Wiley & Sons, Ltd.

ORIGINAL PAPER

Implications of machine–probe combinations for pregnancy dating (CRL, FL, OFD) direction of the ultrasound beam, with the fetus lying in a horizontal position. Much effort has been expended in ensuring that charts and formulae used to date pregnancies are consistent. However, no data are available on the effect of different probe and machine combinations as a factor in the variability of measurements used to date a pregnancy. The aim of this study was to investigate the effect of different ultrasound machine–probe combinations used in routine ultrasound practice, on axial and lateral measurements. The differences that we report do not take into account interobserver variability nor image acquisition variability, which we would expect to have an even greater effect on these measurements. All measurements were based on targets placed at different depths within an ultrasound phantom.

METHODS Sixteen different obstetric ultrasound machine–probe combinations were used during the study. All machines used were compliant with Royal College of Obstetricians and Gynaecologists guidance for anomaly screening and were less than 5 years old4 . The maintenance contract and service record for all devices used in this study were fully up-to-date, including all appropriate software updates. Seven two-dimensional (2D) transabdominal curvilinear probes (probes 1–7), three three-dimensional (3D) transabdominal curvilinear probes (probes 8–10), four 2D transvaginal probes (probes 11–14) and two 3D transvaginal probes (probes 15 and 16) were used with different ultrasound machines. Equipment manufacturers were Aloka (Tokyo, Japan), Siemens (Erlangen, Germany), GE Healthcare (Chalfont St Giles, UK) and Toshiba Medical Systems (Tokyo, Japan). All measurements were taken while imaging the CIRS Model 040, General Purpose Multi-Tissue Ultrasound Phantom (CIRS Inc., Norfolk, Virginia, USA) (Figure 1). This is a commercially available ultrasound phantom designed to mimic the properties of human tissue. A phantom was used in this study to provide a consistent target that was devoid of physiological variability. If the tests had been carried out on a real fetus, it would have been moving 0 cm 1

2 cm

2 1

2

3 4

4

4 cm

5 6

6 cm

Figure 1 Schematic diagram of caliper targets within the CIRS Model 040, General Purpose Multi-Tissue Ultrasound Phantom, showing positions of six vertical filaments for axial measurements and four horizontal filaments for lateral measurements.

Copyright  2012 ISUOG. Published by John Wiley & Sons, Ltd.

195

in the uterus, therefore creating additional errors between measurements. Axial measurements were obtained by imaging vertical filaments 1 to 6 (actual distance, 50 mm) and 3 to 4 (actual distance, 10 mm). Lateral measurements were taken using horizontal filaments 1 to 4 (actual distance, 40 mm) and 2 to 3 (actual distance, 10 mm). Care was taken to ensure that filaments were centered on the probe footprint. Intraoperator variability was calculated from five separate measurements of each frozen image. The mean 40-mm lateral caliper measurements were used to estimate fetal gestational age using standard obstetric dating tables for CRL5 and FL6 and the mean 50-mm axial caliper measurements were used to estimate gestational age using the obstetric dating tables for BPD7 . Results are presented using the mean of the five measurements with 95% CIs to show the uncertainty in the true measurement for each distance for each machine–probe combination. One sample t-tests were used to test if the mean of the observed measurements differed from the true known measurement.

RESULTS Variability of measurement The margin of intraobserver variability was in certain cases greater than the variability that could be ascribed to the mean caliper measurement of different machine–probe combinations. Figure 2 shows the variability for different machine–probe combinations for the axial 10- and 50-mm spaced targets. In all cases intraobserver variability was greater than machine–probe differences for 10-mm spaced targets, but the converse was true for the 50-mm spaced targets. Table 1 shows the measured distances for the axial 10-mm and 50-mm spaced targets using different machine–probe combinations. For the 50-mm spaced targets, most of the measured values were significantly different, whereas only two of the sixteen 10-mm measurements showed statistically significant differences. Figure 2 shows that percentage under and over measurement was consistent at both 10 mm and 50 mm for a given machine–probe combination. For the 10-mm spaced targets the largest mean error was 4% and the smallest was –2% and for the 50-mm spaced targets the largest mean error was 2% and the smallest was –2%. Figure 3 and Table 2 shows the variability for different machine–probe combinations for lateral 10-mm and 40-mm measurements. Machine–probe variability was greater than intraobserver variability for the majority of both the 10-mm and 40-mm target measurements. For the 10-mm spaced targets the largest mean error was 9% and the smallest was –1% and for the 40-mm spaced targets the largest mean error was 6% and the smallest was –2%.

Implications for dating pregnancy The different machine–probe combinations for BPD led to a maximum 4 days’ difference, based on a measured BPD

Ultrasound Obstet Gynecol 2012; 40: 194–199.

Axell et al. 5

0

(b) 5

2

1

3

4

5

7

6

Mean error (%)

(a)

Mean error (%)

196

0

8

9

−5

−5

Machine–probe combination

0

13

12

11

14

(d)

5

Mean error (%)

5

Mean error (%)

Machine–probe combination (c)

10

0

15

16

−5

−5

Machine–probe combination

Machine–probe combination

(a) 12

(b) 12

Mean error (%)

Mean error (%)

Figure 2 Mean error with 95% CI for axial caliper accuracy measurements using: (a) two-dimensional (2D) transabdominal ultrasound; (b) three-dimensional (3D) transabdominal ultrasound; (c) 2D transvaginal ultrasound; (d) 3D transvaginal ultrasound. , 10-mm axial targets; , 50-mm axial targets.

6

0

1

2

−6

3

4

5

6

7

6

0

(d) 12

Mean error (%)

Mean error (%)

6

−6

11

12

13

14

Machine–probe combination

10

9

−6

Machine–probe combination

(c) 12

0

8

Machine–probe combination

6

0

−6

15

16

Machine–probe combination

Figure 3 Mean error with 95% CI for lateral caliper accuracy measurements using: (a) two-dimensional (2D) transabdominal ultrasound; (b) three-dimensional (3D) transabdominal ultrasound; (c) 2D transvaginal ultrasound; (d) 3D transvaginal ultrasound. , 10-mm lateral targets; , 40-mm lateral targets.

Copyright  2012 ISUOG. Published by John Wiley & Sons, Ltd.

Ultrasound Obstet Gynecol 2012; 40: 194–199.

Mean (95% CI) Machine–probe combination 2D-TA 1 2 3 4 5 6 7 3D-TA 8 9 10 2D-TV 11 12 13 14 3D-TV 15 16

10-mm target

50-mm target

9.86 (9.74–9.98)* 9.80 (9.57–10.03) 9.94 (9.73–10.15) 9.90 (9.78–10.02) 9.92 (9.82–10.02) 10.06 (9.91–10.21) 10.40†

48.98 (48.81–49.15)* 49.06 (48.80–49.32)* 49.70 (49.59–49.81)* 49.30† 49.56 (49.48–49.64)* 50.16 (50.08–50.24)* 50.78 (50.62–50.94)*

9.98 (9.88–10.08) 9.90 (9.79–10.01) 9.86 (9.71–10.01)

49.44 (49.36–49.52)* 49.14 (49.01–49.27)* 49.38 (49.34–49.42)*

9.84 (9.76–9.92)* 9.96 (9.77–10.15) 9.88 (9.78–9.98)* 9.86 (9.64–10.08)

49.38 (49.27–49.49)* 49.22 (49.06–49.38)* 49.38 (49.28–49.48)* 50.10 (49.91–50.29)

9.86 (9.68–10.04) 9.90 (9.84–9.96)*

49.00† 49.38 (49.34–49.42)*

*Statistically significantly different (P < 0.05) as CI did not encompass true value. †All five measurements gave the same reading for one machine–probe combination. 2D-TA, two-dimensional transabdominal probe; 2D-TV, two-dimensional transvaginal probe; 3D-TA, three-dimensional transabdominal probe; 3D-TV, three-dimensional transvaginal probe.

Table 2 Values obtained on lateral measurement of targets spaced 10 and 40 mm apart

Machine–probe combination 2D-TA 1 2 3 4 5 6 7 3D-TA 8 9 10 2D-TV 11 12 13 14 3D-TV 15 16

Mean (95% CI) 10-mm target

40-mm target

10.30 (10.21–10.39)* 10.44 (10.32–10.56)* 9.90 (9.79–10.01) 10.12 (10.08–10.16)* 10.28 (10.08–10.48)* 10.34 (10.03–10.65)* 10.49 (10.36–10.62)*

41.22 (41.05–41.39)* 40.68 (40.58–40.78)* 39.06 (38.00–40.12) 40.14 (40.06–40.22)* 40.54 (40.38–40.70)* 41.32 (41.12–41.52)* 41.72 (41.56–41.88)*

10.22 (10.15–10.29)* 10.18 (10.14–10.22)* 10.20 (10.09–10.31)*

40.38 (39.70–41.06) 40.42 (40.35–40.49)* 40.42 (40.32–40.52)*

10.12 (10.02–10.22)* 10.08 (9.98–10.18) 10.20 (10.08–10.32)* 10.86 (10.61–11.11)*

40.54 (40.39–40.69)* 40.36 (40.21–40.51)* 40.74 (40.45–41.03)* 42.40 (41.89–42.91)*

10.22 (10.02–10.42)* 10.28 (10.21–10.35)*

41.08 (40.94–41.22)* 40.98 (40.79–41.17)*

*Statistically significantly different (P < 0.05) as CI did not encompass true value. 2D-TA, two-dimensional transabdominal probe; 2D-TV, two-dimensional transvaginal probe; 3D-TA, three-dimensional transabdominal probe; 3D-TV, three-dimensional transvaginal probe.

Copyright  2012 ISUOG. Published by John Wiley & Sons, Ltd.

8 7 6 5 4 3 2 1 0 −1 −2 −3

197

1

7

8

9

12

14

15

16

Machine–probe combination

Figure 4 Error in gestational age as deduced from differences in measured biparietal diameter, displayed according to machine–probe combination. Data shown for the eight machine–probe combinations associated with greatest error.

Error in gestational age (days)

Table 1 Values obtained on axial measurement of targets spaced 10 and 50 mm apart

Error in gestational age (days)

Implications of machine–probe combinations for pregnancy dating

8 7 6 5 4 3 2 1 0 −1 −2 −3

3

7

8

9

12

14

15

16

Machine–probe combination

Figure 5 Error in gestational age as deduced from differences in measured crown–rump length ( ) and femur length ( ), displayed according to machine–probe combination. Data shown for the eight machine–probe combinations associated with greatest error.

range of 49–51 mm (Table 3 and Figure 4). The impact of different machine–probe combinations on CRL dating was a maximum 2 days’ difference, based on a measured CRL range of 39–42 mm (Table 4 and Figure 5). For FL, the same errors led to a maximum 9 days difference, based on a measured FL range of 39–42 mm (Table 4 and Figure 5).

DISCUSSION The differences in measurements that we report reflect both intraobserver and machine–probe variability. At the larger measurements (40 and 50 mm lateral and axial) the errors are almost exclusively ascribed to differences in machine–probe combinations, while at 10 mm it is clear that the differences are attributable to both intraobserver and machine–probe factors. This is of potential importance, as the differences between machine–probe combinations would lead to clinically significant differences in pregnancy dating depending on gestational age at dating and which parameter is used, for example CRL or FL. The data are likely to be robust, as in all cases a given machine–probe combination either over measured, or under measured, showing consistent errors in the measurement, whose magnitude depended on the size of target

Ultrasound Obstet Gynecol 2012; 40: 194–199.

Axell et al.

198

Table 3 Clinical implications of accuracy of 50-mm axial caliper measurements on estimation of gestational age (GA) based on biparietal diameter (BPD)7 Smallest measurement Ultrasound technique

Largest measurement

Probe

Mean (SD) (mm)

GA (weeks) from BPD

Probe

Mean (SD) (mm)

GA (weeks) from BPD

Difference (days)

1 9 12 15 1 15 1

48.98 (0.19) 49.14 (0.15) 49.22 (0.18) 49.00 (0.00) 48.98 (0.19) 49.00 (0.00) 48.98 (0.19)

20 + 5 20 + 6 20 + 6 20 + 5 20 + 5 20 + 5 20 + 5

7 8 14 16 7 14 7

50.78 (0.08) 49.44 (0.09) 50.10 (0.21) 49.38 (0.04) 50.78 (0.18) 50.10 (0.21) 50.78 (0.18)

21 + 2 20 + 6 21 + 1 20 + 6 21 + 2 21 + 1 21 + 2

4 0 2 1 4 3 4

2D-TA 3D-TA 2D-TV 3D-TV TA TV All

2D, two-dimensional; 3D, three-dimensional; TA, transabdominal; TV, transvaginal. Table 4 Clinical implications of accuracy of 40-mm lateral caliper measurements on estimation of gestational age (GA) based on crown–rump length (CRL)5 and femur length (FL)6

Smallest measurement Ultrasound technique 2D-TA 3D-TA 2D-TV 3D-TV TA TV All

Difference (days) for:

Largest measurement

Probe

Mean (SD) (mm)

GA (weeks) from CRL

GA (weeks) from FL

Probe

Mean (SD) (mm)

GA (weeks) from CRL

GA (weeks) from FL

GA from CRL

GA from FL

3 8 12 16 3 12 3

39.06 (1.21) 40.38 (0.77) 40.36 (0.17) 40.98 (0.22) 39.06 (1.21) 40.36 (0.17) 39.06 (1.21)

10 + 4 10 + 5 10 + 5 10 + 5 10 + 4 10 + 5 10 + 4

22 + 4 23 + 0 23 + 0 23 + 2 22 + 4 23 + 0 22 + 4

7 9 14 15 7 14 14

41.72 (0.18) 40.42 (0.08) 42.40 (0.58) 41.08 (0.16) 41.72 (0.18) 42.40 (0.58) 42.40 (0.58)

10 + 6 10 + 5 10 + 6 10 + 5 10 + 6 10 + 6 10 + 6

23 + 4 23 + 0 23 + 6 23 + 2 23 + 4 23 + 6 23 + 6

2 0 1 0 2 1 2

7 0 6 0 7 6 9

2D, two-dimensional; 3D, three-dimensional; TA, transabdominal; TV, transvaginal.

being measured in the phantom. These findings are likely to be true as one observer performed all the measurements, and intraobserver error was calculated in all cases. The ultrasound phantom provides a consistent target that was devoid of physiological variability, whereas a fetus may be moving in the uterus, thereby creating additional errors between measurements. The phantom weighed exactly the same (4825 g) as specified on the calibration certificate provided by the manufacturer, indicating that there had been no water loss from it, hence the distance between the caliper targets is as specified. Even if there were slight differences in the true distance between the caliper targets within the phantom, this would only introduce a systematic error for all axial and lateral measurements taken using the different ultrasound scanners. This study was designed to make comparisons of the differences in measurement taken using the different scanners, rather than as an evaluation of the absolute measurement size in comparison with the specified caliper target size. Recent guidance recommends that caliper measurements perform to an accuracy appropriate to meet the clinical requirement8,9 . The measurement accuracy of ultrasound machines of the type used in this study is specified by the manufacturer, and ranges from ± 3–5% for linear caliper measurements greater than 20–30 mm to less than 1–1.5 mm for measurements less than

Copyright  2012 ISUOG. Published by John Wiley & Sons, Ltd.

20–30 mm10 – 13 . This is at odds with what is expected clinically3 : a fetal CRL dating measurement normally ranges from 5 to 80 mm with a required accuracy of ± 1 mm. However, for a CRL of 50 mm an ultrasound machine is only actually specified to an accuracy of ± 1.5–2.5 mm, and at 80 mm the specified accuracy is ± 2.4–4 mm. Based on these specifications, the measured CRL using only one machine–probe combination could in fact range from 76 to 84 mm, introducing a difference of 8 mm (10%). This does not take into account errors introduced using other machine–probe combinations. This is of course a far cry from what is expected in our routine clinical practice. Interestingly, for measurements taken using the phantom, 6% of the 10-mm measurements and 31% of the 50-mm measurements exceeded the clinically required limits of accuracy. This is not surprising, as one would expect errors to increase in absolute terms, but reduce in relative terms as the measurement becomes larger. Thus, whereas a measurement error of 1 mm at 20 mm for CRL may be acceptable, at 80 mm the same proportional error (4 mm) would be deemed to be clinically inappropriate. Transforming the differences in these measurement data to commonly used obstetric charts used for dating introduces further differences, depending on which parameter is measured. While we have shown that different machine–probe combinations make a modest

Ultrasound Obstet Gynecol 2012; 40: 194–199.

Implications of machine–probe combinations for pregnancy dating difference to CRL dating (maximum range, 39–42 mm, corresponding to a dating range of 2 days), the same errors resulted in large differences for FL growth (maximum range, 39–42 mm, corresponding to a dating range of 9 days). These absolute measurement differences do not lead to the same dating differences, as incremental CRL growth with time is rapid whereas FL growth is much slower. This leaves the FL measurement particularly prone to both observer and machine–probe errors. However, FL is not normally recommended for pregnancy dating. Though this effect does not appear to have been considered before, the different incremental growth characteristics of CRL, FL and BPD affect their suitability to be used to date pregnancies at different gestational ages. There are a number of limitations that must be taken into consideration when interpreting the outcome of this study. It is normal scientific practice to blind the operator to the device and measurement when performing a reproducibility study. However as with studies performed by Heath et al.14 and Abele et al.15 , it was not possible to blind the operator to which machine–probe combination was being used to take the measurement, as the operator handled the probe and then the controls on the scanner control panel. The operator, although not formally blinded, did not observe the displayed measurement value until both calipers were placed on the target. To increase statistical robustness, we analyzed a mean of five measurements, which is somewhat different from what is done in routine clinical practice, where a single measurement would be taken. A mean value of several measurements is likely to give a more accurate result than is a single random measurement. Interobserver variability and image acquisition variability are not considered within the context of this study, therefore we expect our findings to be an underestimation of the variability in routine clinical practice. A further theoretical consideration occurs in relation to the physical properties of the phantom used to make the measurements. Although the phantom allows for consistent target dimensions in comparison with a moving fetus, the ultrasound beam may behave slightly differently as it passes through the tissue-mimicking material in comparison to the uterus and amniotic fluid. The phantom also incorporates a point-to-point caliper target that is physiologically different from the fetal anatomy when making a CRL or FL measurement. In summary, different machine–probe combinations can produce both statistically and clinically significant differences in measurements, leading to potentially important differences in dating pregnancies. These differences are smallest for small measurements, where intraobserver variability introduces the largest errors. However, as the size of the measurement increases, intraobserver variability makes a progressively less important contribution to measurements than those ascribed to different machine–probe combinations, all of which appear to consistently over or under measure. Though much is written about inter- and intraobserver variability in ultrasound measurements, little thought appears to have been given

Copyright  2012 ISUOG. Published by John Wiley & Sons, Ltd.

199

to the potential for equipment to lead to measurement differences. Clinical guidance for pregnancy dating, and assumptions about ‘reasonable’ clinical measurement differences may need to be revised in the light of these findings. A factor that complicates the practical interpretation of this study is that medical devices are classified differently in the USA and Europe. In the USA ultrasound manufacturers ‘self certify’ for the purposes of Food and Drug Administration/medical device licensing. In Europe ultrasound scanners are classified as Class 2A (medium risk) devices and the process of self certification by manufacturers does not apply for European CE Marking. Nevertheless, manufacturers would be best placed to provide data on the range of measurements expected using their own probe and machine combinations.

REFERENCES 1. Chudleigh T, Thilaganathan B (eds). Obstetric Ultrasound How, Why and When (3rd edn). Elsevier: London, 2004. 2. Altman DG, Chitty LS. New charts for ultrasound dating of pregnancy. Ultrasound Obstet Gynecol 1997; 10: 174–191. 3. Loughna P, Chitty L, Evans T, Chudleigh T. Fetal size and dating: charts recommended for clinical obstetric practice. Ultrasound 2009; 17: 161–167. 4. Royal College of Obstetricians and Gynaecologists (RCOG). Ultrasound Screening: Supplement to Ultrasound Screening for Fetal Anomalies. RCOG Press: London, 2005. 5. Robinson HP, Fleming JE. A critical evaluation of sonar ‘‘crown–rump length’’ measurements. Br J Obstet Gynaecol 1975; 82: 702–710. 6. Chitty LS, Altman DG, Henderson A, Campbell S. Charts of fetal size: 4. Femur Length. Br J Obstet Gynaecol 1994; 101: 132–135. 7. Chitty LS, Altman DG, Henderson A, Campbell S. Charts of fetal size: 2. Head measurements. Br J Obstet Gynaecol 1994; 101: 35–43. 8. Institute of Physics and Engineering in Medicine (IPEM). Report 102. Quality Assurance of Ultrasound Imaging Systems. IPEM Press: York, 2010. 9. Dudley NJ. Revised IPEM guidelines for ultrasound QA. IPEM Physics and Technology of Medical Ultrasound Conference, York, 2009. 10. Aloka. Ultrasound Diagnostic Instrument, prosound α 7, Instruction Manual, Safety Instruction (Volume1/2). Aloka Co, Ltd: Tokyo, 2010. 11. GE Healthcare. Basic User Manual, Voluson E8. Revision 1. GE Healthcare: Milwaukee, WI, USA, 2008. 12. Siemens. ACUSON Antares Ultrasound Imaging System, [1] Instructions for Use. Product version 3.5. Siemens Medical Solution, Inc.: Malvern, PA, USA, 2005. 13. Toshiba Medical Systems. Operation Manual for Diagnostic Ultrasound System Xario XG, Model SSA-680A. Toshiba Medical Systems Corporation: Tokyo, 2010. 14. Heath VC, Southall TR, Souka AP, Novakov A, Nicolaides KH. Cervical length at 23 weeks of gestation: relation to demographic characteristics and previous obstetric history. Ultrasound Obstet Gynecol 1998; 12: 304–311. 15. Abele H, Hoopmann M, Wright D, Hoffmann-Poell B, Huettelmaier M, Pintoffl K, Wallwiener D, Kagan KO. Intra- and interoperator reliability of manual and semi-automated measurement of fetal nuchal translucency by sonographers with different levels of experience. Ultrasound Obstet Gynecol 2010; 36: 417–422.

Ultrasound Obstet Gynecol 2012; 40: 194–199.