DOI:10.2214/AJR.07.2623
AJR 2008; 191:340-345
© American Roentgen Ray Society
Comparison of Fetal Biometric Values with Sonographic and 3D Reconstruction MRI in Term Gestations
Mustapha R. Hatab1,
Michael V. Zaretsky2,
James M. Alexander2 and
Diane M. Twickler2,3
1 Department of Radiology, University of Texas Health Science Center at San
Antonio, MC 7800, 7703 Floyd Curl Dr., San Antonio, TX 78229-3900.
2 Department of Obstetrics and Gynecology, University of Texas Southwestern
Medical Center, Dallas, TX.
3 Department of Radiology, University of Texas Southwestern Medical Center,
Dallas, TX.
Received May 24, 2007;
accepted after revision February 19, 2008.
Address correspondence to M. R. Hatab
(Hatab{at}uthscsa.edu).
Abstract
OBJECTIVE. We sought to compare the fetal biometric values head and
abdominal circumferences, biparietal and occipital–frontal diameters,
and left and right ventricular atrial diameters obtained with contemporaneous
sonography and 3D MRI reconstructions in term pregnancies.
SUBJECTS AND METHODS. A total of 107 nulliparous women evaluated as
having uncomplicated pregnancies and scheduled for induction at 42 completed
weeks gave their informed consent and underwent MRI and sonography within 3
hours of each other. Two single-shot fast spin-echo MRI sequences were
performed with 7- and 4-mm slice thicknesses and no gap. A single observer
performed MRI postprocessing to obtain biometric values. A single sonographer
using a 3- to 5-MHz curvilinear transducer performed transabdominal
sonography. Concordance correlation and Bland-Altman analysis of differences
were performed.
RESULTS. Concordance correlation was poor for both right (0.024) and
left (0.005) ventricular atrial diameters. There were moderate concordance
correlations for head (0.56) and abdominal (0.53) circumferences and
biparietal diameter (0.61). Occipital–frontal diameter had fair
correlation (0.27).
CONCLUSION. Comparison between contemporaneous sonographic and 3D
reconstructed MR images at late gestational ages shows acceptable correlation
between the two techniques for head circumference, abdominal circumference,
and biparietal diameter.
Keywords: biometry • fetal imaging • MRI • obstetrics • sonography
Introduction
The use of MRI as an adjunct to sonography in fetal imaging is becoming
widespread in clinical practice and in the literature. Some investigators
[1–5]
have found a positive effect of the technology on patient care and postnatal
treatment. Fast imaging sequences such as single-shot fast spin echo have all
but eliminated artifacts related to fetal motion while providing superior
resolution and anatomic detail of the fetal brain. A potential advantage of
the improved high contrast resolution is more accurate measurement of fetal
biometric values, which are used as a screening tool to identify fetuses that
fall outside the normal range of measurements and thus are at increased risk
of morphologic abnormalities. Sonography remains the standard for monitoring
of fetal development and biometric analysis. Well-established growth charts
and gestational age estimation tables based on sonographic findings have been
in clinical use since the mid 1970s.
Prenatal MRI is newer than and not as well established as sonography for
evaluating fetal development. Reports
[6–10]
have shown results of direct comparisons of MRI and sonography in the
measurement of fetal biometric values and organ size. As expected, good
agreement between the two methods exists with a slight advantage to MRI in
certain cases, such as fetal weight estimation and measurement of the
posterior fossa [6,
9]. However, because of the
relative youth of the method, MRI-derived fetal measurements to date have been
obtained in relatively small populations. This limitation has resulted in
fairly poor statistical comparison with results of large-population
sonographic biometric studies. We sought to evaluate the agreement in specific
biometric values between contemporaneous sonograms and 3D reconstructed fetal
MR images obtained within 3 hours of each other for fetal head circumference
(HC), abdominal circumference (AC), biparietal diameter (BPD),
occipital–frontal diameter (OFD), and left and right ventricular atrial
diameters in a group of women scheduled for labor induction because of
prolonged pregnancy (42 weeks). The findings should help in assessment of the
validity of sonography-based growth charts for evaluating the normalcy of
MRI-derived biometric values.
Subjects and Methods
This study was ancillary to an investigation aimed at predicting labor
dystocia with MRI [11]. Thus,
all subjects for this study were women with term pregnancies scheduled for
induction at 42 completed weeks. This investigation had institutional review
board approval with only nulliparous women with singleton and uncomplicated
pregnancies invited to enroll. Women with an immediate indication for delivery
were excluded from consideration for the study, as were women with previous
cesarean delivery, hypertension, insulin-dependent diabetes, known fetal
anomalies, or stillbirths. In addition, women weighing more than 360 pounds
(163 kg) were excluded from the study owing to the weight limit of the MRI
table. A total of 107 women (mean age, 22.4 years) participated in the
study.
After informed consent was obtained, sonography was performed with a unit
that had a 3- or 5-MHz curved linear probe (Sequoia, Siemens Medical
Solutions). The same experienced registered sonographer performed studies on
all 107 patients. Within 3 hours of sonography, the patients underwent a
30-minute MRI (1.5-T Signa system, GE Healthcare) examination. The patients
were placed supine in the MRI unit, and a torso surface coil was placed around
the pelvic area centered over the fetal region on all but three of the
patients, whose large size necessitated use of the body coil. A three-plane
rapid localizer acquisition was performed to ensure proper positioning of the
coil and patient. From the localizer acquisition, two 90-second single-shot
fast spin-echo T2-weighted sequences were performed (TE, 60; 44-cm field of
view; 512 x 256 matrix). The first sequence was a 7-mm acquisition
without gap axial to the maternal uterus and incorporating the entire gravid
uterus. The second sequence was a 4-mm acquisition without gap axial to the
maternal pelvis at an angle parallel to the obstetric conjugate and including
the entire fetal head. The entire fetal MRI study was performed in less than
30 minutes.
All MRI data analysis was performed by the same researcher using a 3D
reformatting postprocessing workstation (Advantage Windows AW4.1, GE
Healthcare). AC was the only measurement obtained from the 7-mm acquisitions
(Fig. 1). The 7-mm study was
reconstructed in 3D. The proper orientation showed the curve of the umbilical
vein and the stomach to be visually round. All other biometric
values—BPD, OFD, HC, left and right ventricular atrial diameters,
cisterna magna, and transverse cerebellar diameter—were obtained from
the 4-mm acquisitions. The data set was reconstructed with the 3D capability
of the workstation, and fetal head orientation was manipulated to provide the
optimal slice orientation for biometry. A slice orientation that clearly
showed the thalami and cavum septum pellucidum situated midline in the axial
plane was used to measure BPD, OFD, and HC (Figs.
2A,
2B,
2C, and
2D). For accurate comparison of
the two techniques, biometric values on MRI were measured in the same way as
for a standard antenatal sonographic examination. For example, BPD was
measured from the inner table of the skull on one side to the outer table on
the other side as historically defined with sonography. Postprocessing of all
measurements took less than 20 minutes per patient. The two observers were
blinded to the findings obtained with the other biometric technique. Agreement
between MRI and sonographic biometric values was assessed with Lin's
concordance correlation coefficient
[12,
13] and Bland-Altman analysis
[14]. Lin's concordance
correlation coefficient includes measurements of precision (how far the
observations deviate from the best-fit linear line) and accuracy (how far the
best-fit line deviates from the concordance, or 45° line). The concordance
coefficient of agreement is measured on a scale from 1, which is perfect
agreement, to 0, which is no agreement. This statistical analysis was
considered more descriptive for comparing two methods of measurement than
other procedures, such as paired Student's t tests, least-squares
analysis for slope and intercept, and kappa statistic. Bland-Altman plots are
a graphic representation of the data with the difference between the two
methods plotted against their mean. Bias is the mean difference between the
two methods of measurement and represents systematic error. A 95% CI range
expected to include 95% of the differences between measurements is set at
approximately 2 SD of the mean. SPSS statistical software (version 13, SPSS)
was used to perform statistical calculations; p < 0.05 was
considered significant.
View larger version (129K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2A —21-year-old pregnant woman. Sample raw 4-mm axial MR images
(A–C) and optimized reconstruction (D) for obtaining the
biparietal diameter (BPD), occipital–frontal diameter (OFD), and head
circumference (HC).
|
|
View larger version (139K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2B —21-year-old pregnant woman. Sample raw 4-mm axial MR images
(A–C) and optimized reconstruction (D) for obtaining the
biparietal diameter (BPD), occipital–frontal diameter (OFD), and head
circumference (HC).
|
|
View larger version (171K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2C —21-year-old pregnant woman. Sample raw 4-mm axial MR images
(A–C) and optimized reconstruction (D) for obtaining the
biparietal diameter (BPD), occipital–frontal diameter (OFD), and head
circumference (HC).
|
|
View larger version (106K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2D —21-year-old pregnant woman. Sample raw 4-mm axial MR images
(A–C) and optimized reconstruction (D) for obtaining the
biparietal diameter (BPD), occipital–frontal diameter (OFD), and head
circumference (HC).
|
|
Results
Not all biometric values could be obtained from individual patients with
both techniques (Table 1). The
most obvious deficiency was in imaging the posterior fossa with sonography, as
evidenced by the low numbers for cisterna magna and transverse cerebellar
diameter. Because of such a small data set, no cross-technique comparison was
made for these two biometric values. MRI measurements of BPD, OFD, HC, and AC
were successfully obtained for all 107 subjects. Ventricular atrial diameter
in six patients could not be obtained with MRI owing to excessive motion or
low signal-to-noise ratio.
Table 2 shows the mean
± SD biometric values obtained with both techniques, the Lin's
concordance correlation coefficients, and the 95% confidence limits. Moderate
correlation was found for BPD, HC, and AC; fair correlation for OFD; and poor
correlation for both left and right ventricular atrial diameters. The
correlation is shown in graphic form in Figures
3A,
3B,
4A,
4B,
5A,
5B,
6A, and
6B. The best-fit line for AC
was y = 0.96x + 34.6, for BPD was y = 1.1x
– 7.4, for HC was y = 1.1x – 39.4, and for OFD
was y = 0.36x + 74.2. Results of Bland-Altman analysis of
differences showed bias of 3% of the mean sonographic value for BPD, 4% for
OFD, 3% for HC, and 6% for AC. Table
2 and Figures 3A,
3B,
4A,
4B,
5A,
5B,
6A, and
6B show that the MRI
measurements of BPD, OFD, HC, and AC in fetuses at 40 weeks of gestation and
beyond were significantly greater than the sonographic values; the right and
left ventricular atrial MRI measurements were significantly less than the
sonographic values.
View this table:
[in this window]
[in a new window]
|
TABLE 2: Biometric Values and Lin's Concordance Correlation Coefficient with
Lower and Upper 95% Confidence Limits
|
|
Discussion
Sonography has been and will most likely remain the primary imaging
technique for routine antenatal examinations. A long history of successful
clinical use has solidified its role in examination of developing fetuses and
has resulted in accurate fetal growth charts. Templates of normal growth
derived from fetal MRI examinations are not available owing to the relative
youth of the method and consequently the smaller cohorts. Use of MRI to study
the fetal brain is gaining widespread acceptance because of improved spatial
resolution and soft-tissue contrast compared with sonography
[2,
15,
16]. However, routine biometry
of the fetal brain and cranial structures with MRI is not typically performed.
It is current practice to check biometric values obtained with MRI against
sonographically derived normal growth charts. The reasoning is that the
biometric values obtained, BPD and OFD for example, should be consistent
regardless of the technique used to obtain them. In this study, we sought to
validate this claim by comparing biometric values obtained at term with
contemporaneous sonography and MRI of a group of 107 patients.
Few reports have presented complete cerebral biometric data obtained with
MRI. Twickler et al. [17]
derived a linear regression equation relating the cisterna magna measurement
to gestational age based on results for 23 patients. Reichel et al.
[6] were the first to compare
established fetal gestational age with BPD, HC, and cerebellar width. Good
correlation was observed, but the age estimates from MRI-derived biometric
values were obtained from the widely accepted, sonographically derived Hadlock
et al. [18] and Hill et al.
[19] tables. Reichel et al.
suggest that "MR biometry needs to have its own nomograms
established" [6].
Garel [20] departed from
sonographically defined views for cerebral biometry. For example, Garel
provided two measurements for BPD: a bone measurement defined as the distance
between the two internal tables of the skull and a cerebral measurement
defined as the maximum transverse diameter of the brain. Both of these
measurements were obtained from coronal views and not from the typical
transthalamic axial view. With these optimized MRI orientations, Garel
[21,
22], from a cohort of 225
fetuses, has provided the only, to our knowledge, MRI-derived fetal biometric
growth charts. In all previous studies to our knowledge, however, fetuses have
not been a cohort with truly normal values because MR images were obtained for
indications other than CNS abnormalities.
In this study, we found that axial-plane optimization can be achieved with
a relatively effortless postprocessing technique from single 90-second
acquisitions in a term fetus. The observed correlations between MRI and
sonographic measurements of BPD, HC, and AC suggest that sonographic templates
from a large cohort can be used in defining normal MRI-measured growth. The
mean differences, which represent bias or systematic difference between the
two methods of measurement, for these three biometric values were less than 6%
of their mean sonographic values. In addition, these differences were less
than twice the SD around the mean of the corresponding sonographic
measurements, allowing for the inherent limitations of individual sonographic
biometry values at term. For OFD, the means of the two methods differed only
4%, but the correlation was low. One possible explanation for this finding is
that the fetal lie in these term pregnancies affected the utility of
sonography in evaluation of certain biometric values with a high degree of
accuracy, especially because OFD is not routinely measured at antenatal
sonographic examinations. As for the ventricular atrial measurements, the poor
correlation can be attributed to errors introduced in reconstruction and
postprocessing; namely, when 4-mm slices are reconstructed and used to measure
a 5- to 7-mm structure, the chances of introducing error into measurement
increase. Because transverse measurements of the atria are commonly used as a
reference for defining ventriculomegaly, such measurements should be obtained
while the patient is undergoing MRI with the proper slice orientation and not
later during postprocessing and study reconstruction.
An expected finding of this study that reinforces the advantage of MRI over
sonography for imaging the posterior fossa at term is shown in
Table 1. With sonography, it
was possible to measure the transverse cerebellar diameter and the cisterna
magna in only 10 of the 107 fetuses at term. This limitation is to be
expected, especially because of the late gestational age of the population,
which would result in the most pronounced attenuation of the ultrasound beam
in traversing the skull. In addition, that head measurements during routine
sonography could not be obtained in nearly 10% of our patients is likely due
to the low position of the fetal head in the pelvis at term. This
configuration makes it difficult to obtain the proper imaging plane for
sonography but is not an issue with MRI. The merits of using the historical
orientations to obtain cerebral biometric values versus some newly defined
MRI-optimized orientations will have to be evaluated before final
recommendations can be made.
During investigations for possible CNS abnormalities, the fetal brain is
typically imaged with MRI in the three orthogonal planes with thin slices, and
single slices with the most optimal orientation are acquired to help with the
diagnosis. In this study, however, all of the patients had uncomplicated
pregnancies with no CNS problems judged with sonography. A specific aim of
this study was to show that by using a quick 90-second MRI acquisition of the
whole fetus or fetal brain, it is feasible to obtain biometric values from the
reconstructed images off-line. We believe that obtaining measurements from 3D
reconstructed volumes is an alternative to direct measurement for certain
fetal biometric values, especially when thin-enough slices (such as our 4-mm
slices) are used. We chose 3D reconstruction and not direct plane measurement
in this CNS-normal population because our main concern was examination time.
Time is always of the essence in fetal MRI examinations. The patient may be
uncomfortable on the table, which increases the likelihood of motion with
increasing imaging time. More important, when much time is spent trying to
find the proper orientation in which to image a fetus, the likelihood of fetal
motion increases. Thus, with a quick 90-second acquisition and postprocessing
to obtain measurements, the risk of motion artifact is reduced compared with
acquisition of multiple individual 20-second images in an attempt to locate
the best orientation in each of the three axes.
Some limitations of this study should be addressed. First, the subjects
were enrolled in a study to evaluate the efficacy of MRI in prediction of
dystocia; thus all were recruited from the postterm clinic. This bias in
gestational age favors MRI when measurements of the posterior fossa are
attempted. It would have been preferable to perform side-by-side comparisons
between the two techniques throughout gestation, thus overcoming the
impairment imposed by attenuation of the ultrasound beam through bone at late
gestation, but an extremely large cohort would have been needed to achieve
acceptable statistical power. The other repercussion of late gestational age
is the scarcity of biometric values in the literature for comparison.
The second limitation was that the sample size was a relatively small 107.
However, the population was the largest of which we are aware to be examined
with both sonography and MRI within 3 hours of each other. The third
limitation was that the reliability of the measurements and interobserver
error were not quantified. However, both the sonographers and the MRI
operators were highly experienced in obtaining fetal biometric values and thus
most likely provided accurate data. It is likely that the 3–6%
differences between the two methods were due to interobserver and
intraobserver variability, especially with the observed systematically larger
measurements of the majority of biometric values measured with MRI compared
with sonography. Finally, although a classic biometric value, femur length was
not included in this study because with MRI the signal void of the bony cortex
did not allow reliable measurements. The orientation of the femur is much less
predictable than those of the head and abdomen, rendering the postprocessing
technique more time-consuming and less accurate.
The results of this first, to our knowledge, direct comparison of normal
fetal biometric data obtained with contemporaneous sonography and 3D
reconstructed MRI at late gestational age show acceptable correlation between
the two techniques for certain biometric values.
References
- Glenn A, Barkovich J. Magnetic resonance imaging of the fetal brain
and spine: an increasingly important tool in prenatal diagnosis. Am
J Neuroradiol 2006; 27:1807
-1814[Abstract/Free Full Text]
- Twickler DM, Magee KP, Caire J, Zaretsky M, Fleckenstein JL, Ramus
RM. Second opinion magnetic resonance imaging for suspected fetal central
nervous system abnormalities. Am J Obstet Gynecol2003; 188:492
-496[CrossRef][Medline]
- Miller E, Ben-Sira L, Constantini S, Beni-Adani L. Impact of
prenatal magnetic resonance imaging on postnatal neurosurgical treatment.
J Neurosurg 2006;105
: 203-209[Medline]
- Simon EM, Goldstein RB, Coakley FV, et al. Fast MR imaging of fetal
CNS anomalies in utero. Am J Neuroradiol2002; 21:1688
-1698
- Whitby EH, Paley MN, Sprigg A, et al. Comparison of ultrasound and
magnetic resonance imaging in 100 singleton pregnancies with suspected brain
abnormalities. BJOG 2004;111
: 784-792[CrossRef][Medline]
- Reichel TF, Tamus RM, Caire JT, Hynan LS, Magee KP, Twickler DM.
Fetal central nervous system biometry on MR imaging.
AJR 2003; 180:1155
-1158[Abstract/Free Full Text]
- Duncan KR, Issa B, Moore R, Baker PN, Johnson IR, Gowland PA. A
comparison of fetal organ measurements by echo-planar magnetic resonance
imaging and ultrasound. BJOG 2005;112
: 43-49[Medline]
- Garel C, Alberti C. Coronal measurements of the fetal lateral
ventricles: comparison between ultrasonography and magnetic resonance imaging.
Ultrasound Obstet Gynecol 2006;27
: 23-27[CrossRef][Medline]
- Zaretsky MV, Reichel TF, McIntire DD, Twickler DM. Comparison of
magnetic resonance imaging to ultrasound in the estimation of birth weight at
term. Am J Obstet Gynecol 2003;189
: 1017-1020[CrossRef][Medline]
- Hassibi S, Farhataziz N, Zaretsky M, McIntire D, Twickler DM.
Optimization of fetal weight: estimates using MRI. AJR2004; 183:487
-492[Abstract/Free Full Text]
- Zaretsky MV, Alexander JM, McIntire DD, Hatab MR, Twickler DM,
Leveno KJ. Magnetic resonance imaging pelvimetry and the prediction of labor
dystocia. Obstet Gynecol 2005;106
: 919-926[Medline]
- Lin LI. A concordance correlation coefficient to evaluate
reproducibility. Biometrics 1989;45
: 255-268[CrossRef][Medline]
- Lin LI. A note on the concordance correlation coefficient.
Biometrics 2000;56
: 324-325[CrossRef]
- Bland JM, Altman DG. Statistical methods for assessing agreement
between two methods of clinical measurement. Lancet1986; 1:307
-310[CrossRef][Medline]
- Garel C. New advances in fetal MR neuroimaging. Pediatr
Radiol 2006; 36:621
-625[CrossRef][Medline]
- Coakley FV, Glenn OA, Qayyum A, Barkovich AJ, Goldstein R, Filly
RA. Fetal MRI: a developing technique for the developing patient.
AJR 2004; 182:243
-252[Free Full Text]
- Twickler DM, Reichel T, McIntire DD, Magee KP, Ramus RM. Fetal
central nervous system ventricle and cisterna magna measurements by magnetic
resonance imaging. Am J Obstet Gynecol2002; 187:927
-931[CrossRef][Medline]
- Hadlock FP, Deter RL, Harrist RB, Park SK. Estimating fetal age:
computer-assisted analysis of multiple fetal growth parameters.
Radiology 1984;152
: 497-501[Abstract/Free Full Text]
- Hill LM, Guzick D, Fries J, Hixson J, Rivello D. The transverse
cerebellar diameter in estimating gestational age in the large for gestational
age fetus. Obstet Gynecol 1990;75
: 981-985[Medline]
- Garel C. Fetal cerebral biometry: normal parenchymal findings and
ventricular size. Eur Radiol 2005;15
: 809-813[CrossRef][Medline]
- Garel C. Methodology and results. In: Garel C, ed. MRI
of the fetal brain: normal development and cerebral pathologies.
New York, NY: Springer, 2004
- Garel C. The role of MRI in the evaluation of the fetal brain with
an emphasis on biometry, gyration and parenchyma. Pediatr
Radiol 2004; 34:694
-699[Medline]
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
Top
Abstract
Introduction
