European Journal of Radiology 74 (2010) e107–e111

Contents lists available at ScienceDirect

European Journal of Radiology

journal homepage: www.elsevier.com/locate/ejrad

Pelvimetry revisited: Analyzing cephalopelvic disproportion

Miriam S. Lenhard a,1 , Thorsten R.C. Johnson b,∗,1 , Sabine Weckbach b ,
Konstantin Nikolaou b , Klaus Friese a , Uwe Hasbargen a
a
Department of Obstetrics and Gynecology, Ludwig-Maximilians-University of Munich, 81377 Munich, Germany
b
Department of Radiology, Ludwig-Maximilians-University of Munich, Marchioninistrasse 15, 81377 Munich, Germany

a r t i c l e i n f o a b s t r a c t

Article history: The objective of this study was to assess the clinical value of pelvimetry to predict dystocia due to
Received 6 January 2009 cephalopelvic disproportion.
Accepted 15 April 2009 63 patients who had received an abdominal CT scan postpartum were included. Pelvimetry was
performed retrospectively with these datasets on a 3D workstation; there were no CT examinations
Keywords: performed solely for pelvimetry, and there was no radiation exposure for study purposes. Patients
Pelvimetry
were divided into three groups by the course of birth, i.e. normal vaginal delivery (A), dystocia due to
Cephalopelvic disproportion
cephalopelvic disproportion (B) and other patients (C). Previously described methods were evaluated for
Computed tomography
their accuracy in diagnosing cephalopelvic disproportion.
The pelvimetric parameters did not show significant differences between groups A (n = 20) and B
(n = 20) except for the sagittal mid-pelvic diameter (q) with 12.7 ± 0.6 cm vs. 11.9 ± 0.6 cm (p = 0.0001).
The ROC analysis of the previously described methods showed areas under the curve between 0.50
and 0.67. The ROC curves for q had an area of 0.88, providing 85% sensitivity with 85% speci-
ficity.
In conclusion, the sagittal mid-pelvic diameter shows potential to detect cephalopelvic disproportion
with acceptable accuracy. With the information gained on the CT data, a prospective trial based on MR
imaging can be set up to validate the diagnostic accuracy.
© 2009 Elsevier Ireland Ltd. All rights reserved.

1. Introduction pregnancy to avoid radiation exposure of the fetus [2,4–6]. For

Prolonged delivery is associated with increased fetal and mater-
nal morbidity and mortality. Cephalopelvic disproportion is the
most important cause of prolonged birth, baring risks of difficult
or even unsuccessful vaginal births with the necessity of vacuum
extraction or forceps delivery [1]. Therefore, it is important to iden-
tify those women who are at risk for dystocia [2] and to choose
the most appropriate way of delivery at an early stage in preg-
nancy. Altogether, little progress has been made in diagnosing
cephalopelvic dysproportion. Many studies reported very limited
diagnostic accuracy of various pelvimetric indices in the prediction
of dystocia. In fact, many authors regard pelvimetry as an unsuc-
cessful and dispensible means to attempt to predict cephalopelvic
disproportion. The lost interest in pelvimetry has been aroused
since an accurate pelvimetry on cross-sectional images without
ionizing radiation has become possible by magnet resonance imag-
ing (MRI) [3], which is indeed very important for pelvimetry during
∗ Corresponding author. Tel.: +49 89 7095 3620.
E-mail address: thorsten.johnson@med.uni-muenchen.de (T.R.C. Johnson).
1
Co-first author, both authors contributed equally to this article.
conventional imaging, measurement errors around 10% have been
described, whereas MRI seems to be quite accurate with errors
around 1% [3]. A major problem in MRI with conventional sequences
is the necessity to accurately plan the scans on preview images,
because the cross-sectional images are restricted to the imag-
ing planes and a retrospective adjustment is not possible based
on the once-acquired data. Despite of the problematic issue of
ionizing radiation, CT has been described to be an accurate modal-
ity for pelvimetry [7,8]. As CT is routinely performed as a spiral
data acquisition with continuous volume coverage, a retrospective
pelvimetry is possible even with CT data acquired for other indi-
cations. With three-dimensional volume rendering techniques, a
precise measurement of pelvic dimensions is possible by inter-
actively turning and cutting the dataset. Our previous feasibility
study has shown a good accuracy with high reproducibility and
low inter-observer variability even on 5 mm axial slices. Thus,
the CT data of patients in whom the course of birth is known
can be used to analyze the value of pelvimetry for the assess-
ment of cephalopelvic disproportion. The aim of this study was
to assess the diagnostic accuracy of pelvimetry in the identifica-
tion of women at risk for dystocia based on a retrospective CT data
evaluation.

0720-048X/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.ejrad.2009.04.042
e108 M.S. Lenhard et al. / European Journal of Radiology 74 (2010) e107–e111

2. Materials and methods
2.1. Available CT data
The picture archiving and communication system (PACS) of a
university hospital was screened for abdominal CT scans of patients
referred from the obstetric ward during the last 6 years. 63 patients
who had received a CT scan within 1 week postpartum between
01/2000 and 05/2008 were identified and included in the analy-
sis. Indications for CT were suspicion of ovarian vein thrombosis
(in 33 patients), abscess (4) or hemorrhage (3), fever (2) or inflam-
matory laboratory parameters of unknown cause (6) or persistent
abdominal pain (15). The retrospective evaluation is warranted by
the university’s ethics committee guidelines and adhered to the
World Medical Association Declaration of Helsinki.
All CT examinations had been acquired on spiral CT scanners
with routine abdominal protocols in venous phase after intravenous
contrast material application. Depending on the year of acquisition,
4, 16 or 64 slice CT scanners (Siemens Somatom Volume Zoom, Sen-
sation 4, Sensation 16 or Sensation 64) had been used. Images had
been reconstructed at 5 mm slice thickness and increment with a
standard medium-soft convolution kernel. For 15 exams acquired
on the 64 slice system, 1 mm slices were available. All exams had
been archived in the long-term storage of the PACS system. There
were no CT examinations performed solely for pelvimetry, and
there was no radiation exposure for study purposes.

Fig. 1. Volume rendered images of the bony pelvis in a superior–anterior (a) and a posterior view (b) and a lateral view of the left half of the pelvis cut in strict sagittal
direction (c). The lines indicate the measurement of the transverse diameter of the inlet (a), the interspinous and intertuberous distance (upper and lower line in b), of the
obstetric conjugate, the sagittal mid-pelvic and outlet diameter (top to bottom lines in c).
 M.S. Lenhard et al. / European Journal of Radiology 74 (2010) e107–e111
Table 1
Maternal and fetal demographic data for groups A and B.
Demographic data Normal delivery group (A, n = 20)
Maternal
Age (years) 33 ± 6 32 ±
Primipara 5 15
Height (cm) 167 ± 6 165 ±
Weight at early pregnancy (kg) 66.4 ± 9.2
Body mass index (kg/m2 ) 26.3 ± 4.3
Weight gain (kg) 16.3 ± 7.4
Fetal
Birth weight (g) 3301 ± 720
Body length (cm) 50.0 ± 4.4
Head circumference (cm) 33.3 ± 4.0
pH of umbilical cord blood 7.3 ± 0.1
“p” by t-test except * Chi-square test.
2.2. Pelvimetry
Based on these images, pelvimetry was performed retrospec-
tively on a 3D workstation (Siemens Multi-Modality Workplace,
Siemens, Forchheim/Germany) using volume rendered images by
two independent blinded observers as previously reported [8]. The
technique implies measurements in standard cranial, posterior and
lateral views: the transverse diameter of the inner pelvis was mea-
sured on the 3D image from a cranial view (Fig. 1a). On a posterior
view, the interspinous distance was measured as the shortest dis-
tance between both ischial spines and the intertuberous distance
was masured as the widest distance between the ischial tuberosi-
ties (Fig. 1b). Then, the dataset was cut in mid-sagittal direction
and sagittal measurements were obtained on a strict lateral view.
These included the obstetric conjugate as the shortest distance
between promontory and symphysis, the sagittal outlet as the dis-
tance between the inferior inner aspect of the symphysis and the
distal end of the sacrum, and the mid-pelvic sagittal diameter from
the inferior inner aspect of the symphysis along the plane of the
spinous process to the sacrum (Fig. 1c). All measurements were
obtained independently by both observers, and inter-observer vari-
ability was quantified using the Bland–Altmann method.
2.3. Correlation to clinical data
Medical records, clinical charts including birth protocols or
surgery reports were reviewed to analyze maternal and infant data
(cf. Table 1). Maternal clinical data included age, height, weight,
BMI, parity, mode of delivery, indication for intervention during
delivery, anesthetics, postpartum complications, indication and
diagnosis of postpartum CT scan. Infant data included birth weight,
length, head circumference, umbilical cord pH and Apgar scores
at 1, 5 and 10 min. To take other mechanical factors than pelvic
dimensions into account which can have an effect on dystocia, birth
position, presentation, flexion attitude and lie were recorded from
the patient files. In addition, complications involving the umbil-
ical cord including nuchal cord, cord knot or cord prolapse were
assessed.
Cephalopelvic disproportion was diagnosed if there was full cer-
vical dilatation and adequate contractions of the uterus but no
progress in labor for 2 h, and an intervention by forceps delivery,
vacuum extraction or cesarean section was initiated.
Based on this data, patients were divided into three groups by
the course of birth, i.e. normal spontaneous vaginal delivery after
the 38th gestational week (A), assisted birth for dystocia due to
cephalopelvic disproportion (B) and patients who did not match
either group (C). The latter included patients with elective cesarean
section, multiple pregnancy or delivery before the 38th gestational
week, so that definite conclusions on the influence of pelvimetric
e109
Dystocia group (B, n = 20) p
6 0.838
0.004*
6 0.367
73.0 ± 23.4 0.274
28.1 ± 8.6 0.440
12.9 ± 4.5 0.100
3269 ± 565 0.888
51.4 ± 2.9 0.262
35.9 ± 3.9 0.144
7.3 ± 0.1 0.149
dimensions were not possible, even though some of these patients
did go through labor. Group C was therefore excluded from further
analysis.
2.4. Assessment of diagnostic accuracy for cephalopelvic
disproportion
Various previously described methods used to diagnose
cephalopelvic disproportion were evaluated in a blinded man-
ner for their accuracy in diagnosing cephalopelvic disproportion.
These included the methods described by Colcher [9,10], Mengert
[11], Borell [12], Friedman [13] and the fetal–pelvic index [14].
The respective index was calculated for each patient to determine
whether disproportion would have been diagnosed correctly or not.
The diagnostic accuracy was analyzed by ROC curves representing
sensitivity and specificity of the method including 95% confidence
intervals.
2.5. Statistical analysis
Kolmogorov–Smirnov tests were applied to verify normal dis-
tribution of pelvimetric variables. Continuous variables are given
as mean ± standard deviation. Differences between pelvimetric
dimensions were tested for their significance using the Student’s
t-test. To analyze the significance of differences between paramet-
rical data, the Chi-square test was applied. Statistical significance
was assumed if p-values were less or equal to 0.05. Diagnostic accu-
racy was calculated based on standard contingency tables. Receiver
operating (ROC) curves were calculated based on the results of
the respective diagnostic method for each patient with reference
to the clinical outcome. Inter observer variability was assessed
using Bland–Altmann analysis [15] including mean differences,
standard deviation and limits of agreement. All statistical tests were
performed with MedCalc (Version 7.3.0.1, MedCalc Software, Mari-
akerke, Belgium).
3. Results
63 patients referred from the obstetric ward for abdominal CT
scans were identified. The review of the patient records revealed 20
normal vaginal deliveries (group A). Dystocia was suspected and
intervention initiated in another 20 patients (group B), i.e. sec-
ondary cesarean section in 15, vacuum extraction in 5 cases. In
the remaining 23 patients (group C), there was no clear evidence
whether there may have been disproportion or not, e.g. because
a cesarean section was initiated for other reasons than dystocia
(multiple pregnancy, prematurity).
Regarding clinical parameters including umbilical cord artery
pH value, fetal weight and fetal length as well as mother’s weight,
e110 M.S. Lenhard et al. / European Journal of Radiology 74 (2010) e107–e111

Table 2
Pelvimetric data for both patient groups.

Pelvimetric data (cm) Normal delivery group (A, n = 20) Dystocia group (B, n = 20) p

m (obstetric conjugate) 12.0 ± 0.9 12.0 ± 1.2 0.8922

c (transverse inlet) 12.9 ± 0.9 12.7 ± 1.1 0.5298
q (sagittal interspinal) 12.7 ± 0.6 11.9 ± 0.6 0.0001*
s1 (sagittal outlet) 12.1 ± 0.9 11.5 ± 0.8 0.2696
h (interspinal) 10.9 ± 0.7 10.6 ± 0.6 0.1867
I (intertuberous) 12.0 ± 0.9 11.7 ± 0.8 0.2920
*
Significant difference.

height and body mass index, Apgar values or birth injuries, there
were no statistically significant differences between both groups
except for the number of primipara which was significantly higher
in group B (Table 1). In group A there were only cephalic presenta-
tions recorded, 19 in first occipito-anterior presentation, 1 in second
occipito-anterior presentation. In group B, there were two cases of
high stage longitudinal position. Apart from that, there were only
cephalic presentations recorded, 17 in first occipito-anterior and
1 in second occipito-anterior presentation. Regarding the umbili-
cal cord there were no remarkable findings except for one case of
double nuchal cord entanglement in group B, in which a vacuum
extraction was applied.
Differentiating groups A (normal vaginal delivery) and B
(dystocia due to disproportion), the only significantly different
pelvimetric parameter was the sagittal mid-pelvic diameter at the
level of the spinous processes (q) with 12.7 ± 0.6 cm vs. 11.9 ± 0.6 cm
(p = 0.0001). The other pelvimetric parameters did not show signif- Fig. 2. ROC analysis showing the diagnostic accuracy of various tests in the predic-
tion of cephalopelvic disproportion. The sagittal mid-pelvic diameter q and the same
icant differences (cf. Table 2). Regarding the two cases of high stage
value divided by the fetal head circumference achieve the highest areas under the
longitudinal position and one double nuchal cord entanglement in curve.
group B, all three women had low sagittal mid-pelvic dimensions
(11.3, 12.0 and 11.9 cm), respectively.
Overall inter-observer agreement showed a mean difference Relating the sagittal mid-pelvic diameter to the fetal head cir-
of −0.18%, a standard deviation 2.75% and limits of agree- cumference (q/HC), a similar diagnostic accuracy can be obtained
ment at −5.58% and +5.22%; correlation coefficient was 0.95. with an area under the curve of 0.859, providing 70% sensitivity
Regarding the different pelvimetric dimensions, the correlation with 90% specificity at a cut-off value of 0.34 (relation q/HC).
coefficients were very high for obstetric conjugate (0.96), trans-
verse inlet (0.99), sagittal mid-pelvic diameter (0.96), sagittal 4. Discussion
outlet (0.96) and limited for interspinal and intertuberous distance
(both 0.91). The pelvimetric dimensions we obtained in our study group are
The calculation of the previously described indices for the pre- generally in good agreement with published data [16,17], although
diction of cephalopelvic disproportion yielded very variable results. somewhat different from older values, which may be attributable
Fig. 2 shows the diagnostic accuracy of the methods in ROC curves. to a general increase of average body size over the last century
Table 3 lists the areas under the ROC curves including 95% confi- [18]. As the CT scans had been obtained within 1 week after deliv-
dence intervals, standard error, sensitivity and specificity at highest ery, it is unlikely that the pelvic dimensions analyzed in our study
diagnostic accuracy. would have changed significantly from the situation during birth.
The values range from 0.50 to 0.67, and none of the 95% confi- The patient groups were well balanced for most demographic fac-
dence intervals excludes 0.5, which indicates that the tests are of tors. The exception that we observed a higher rate of primipara in
very questionable value in our patient group. the ‘dystocia’ group B can be attributed to the bias caused by the fact
Regarding the mere pelvimetric measurements, the ROC curve that most women with a history of previous dystocia will deliver by
for q has an area under the curve of 0.871, with a 95% confidence cesarean section at subsequent births. Also, the rate of dystocia is
interval of 0.726–0.957. Thus, the determination of q alone offers very high in our study group because these patients are much more
a sensitivity of 85% with 85% specificity for the identification of likely to require a CT scan postpartum than those with a normal
disproportion if a threshold of 12.1 cm is applied (Fig. 3). vaginal delivery.

Table 3
ROC analysis of the diagnostic accuracy of pelvimetric indices.

Index Area under curve 95% confidence interval Standard error Sensitivity/specificity at
maximal diagnostic accuracy

Borell 0.500 0.336–0.664 0.049 0 0

Colcher 0.524 0.358–0.686 0.093 10.0 95.0
Fetal–pelvic index 0.670 0.501–0.811 0.087 55.0 75.0
Friedman 0.595 0.426–0.749 0.091 40.0 78.9
Mengert 0.501 0.337–0.665 0.094 95.0 5.0
q (≤12.1 cm) 0.871 0.726–0.957 0.059 85.0 85.0
q/HC (≤0.34) 0.859 0.710–0.949 0.062 70.0 89.5
 M.S. Lenhard et al. / European Journal of Radiology 74 (2010) e107–e111
Fig. 3. Box-and-Whisker plot of the sagittal mid-pelvic diameter for patient groups
A and B showing the good separation with little overlap at a threshold of 12.1 cm.
The observation that the sagittal diameter of the mid-pelvis has
high importance is in agreement with Zaretsky et al. who made very
similar observations in MRI pelvimetric data [17]. It seems that the
mid-pelvis, through which the fetus’ head normally passes in sagit-
tal orientation, is just the mechanically narrowest part of the human
birth canal. Other factors including weight of mother and fetus seem
to be less important. The three cases of nuchal cord entanglement
or high stage longitudinal position in group B showed low q values,
which does not allow to draw conclusions on the effect on these
factors. Actually, the pelvic dimensions may have represented the
initial cause of the high stage longitudinal positions as they may
prevent appropriate rotation of the fetal head. Still, it is not surpris-
ing that our observation reports a limited accuracy of pelvimetry in
the prediction of dystocia, because it cannot take all relevant fac-
tors into account. The uterine activity is not considered although
it represents an important factor of the delivery process. The size
of the fetus, as another factor, can be taken into account by relat-
ing the pelvic dimensions to the head circumference or diameter,
which may be useful if the size of the fetal head differs significantly
from average at term.
The other pelvimetric indices for the prediction of cephalopelvic
disproportion which we applied to our data, including the ones
described to predict by Borell et al., Colcher et al., Friedman et al.
and Mengert et al., showed a rather low diagnostic accuracy. Their
low clinical value is reflected by the fact that pelvimetry has been
abandoned at most centers. Our observation that the sagittal mid-
pelvic diameter can predict disproportion more reliably may help
to revive the method and improve patient outcomes.
Our study was performed as a retrospective analysis based on
CT data acquired for postpartum complications. The reason for the
choice of this modality is the availability of data with continuous z-
e111
axis coverage, which is not given in standard pelvic MRI. However,
this is not to imply that CT is ideal for the purpose of pelvime-
try. Even though the radiation can be reduced significantly with
exams at extremely low tube current, the avoidance of radiation
exposure for the fetus would of course make MRI the modality of
choice in pregnancy. With the information gained on the CT data in
this retrospective analysis, a prospective trial based on MR imaging
can be set up to validate the diagnostic accuracy in the prediction
of cephalopelvic disproportion. New MR imaging techniques with
continuous volume coverage such as VIBE (volumetric interpolated
breath-hold examination) [19] may be useful to make multiplanar
reformats possible.
5. Conclusion
Previously reported methods to predict cephalopelvic dispro-
portion did not perform well in our study population. However,
the sagittal mid-pelvic diameter at the level of the spinous process
shows potential to detect disproportion with acceptable accuracy.
Thus, this index may offer a higher diagnostic accuracy and make
the method more reliable. Prospective MRI studies are required to
validate these results and assess the predictive value.
References
[1] McMahon MJ, Luther ER, Bowes Jr WA, Olshan AF. Comparison of a trial of labor
with an elective second cesarean section. N Engl J Med 1996;335(10):689–95.
[2] Ferguson 2nd JE, Sistrom CL. Can fetal–pelvic disproportion be predicted. Clin
Obstet Gynecol 2000;43(2):247–64.
[3] Stark DD, McCarthy SM, Filly RA, Parer JT, Hricak H, Callen PW. Pelvimetry by
magnetic resonance imaging. AJR 1985;144(5):947–50.
[4] Federle MP, Cohen HA, Rosenwein MF, Brant-Zawadzki MN, Cann CE. Pelvime-
try by digital radiography: a low-dose examination. Radiology 1982;143(3):
733–5.
[5] van Loon AJ, Mantingh A, Serlier EK, Kroon G, Mooyaart EL, Huisjes HJ.
Randomised controlled trial of magnetic-resonance pelvimetry in breech pre-
sentation at term. Lancet 1997;350(9094):1799–804.
[6] Levine D, Barnes PD, Edelman RR. Obstetric MR imaging. Radiology
1999;211(3):609–17.
[7] Breidt D, Bouquigny F, Clement JP, et al. Helical CT pelvimetry: advantages of a
low dose volume acquisition technique. J Radiol 2003;84(9):1027–30.
[8] Balleyguier C, Jouanic JM, Correas JM, Benachi A, Dumez Y, Menu Y. CT pelvime-
try: a new approach using multi detector CT and volume rendering. J Radiol
2003;84(4 Pt 1):425–7.
[9] Brown RC. A modification of the Colcher–Sussman technique of X-ray pelvime-
try. Am J Roentgenol Radium Ther Nucl Med 1972;115(3):623–9.
[10] Freeman DW. X-ray pelvimetry by the Colcher–Sussman method; a study com-
paring it with stereoscopic parallax pelvimetry. Minn Med 1956;39(9):583–5,
passim.
[11] Mengert WF, Korkmas MV. 3,772 labors following radiographic mensuration.
Am J Obstet Gynecol 1957;74(1):151–8.
[12] Borell U. Orthodiagraphic pelvimetry with special reference to capacity of distal
part of pelvis and pelvic outlet. Acta Radiol: Diagn 1964;2:273–82.
[13] Friedman EA, Taylor MB. A modified nomographic aid for X-ray cephalopelvime-
try. Am J Obstet Gynecol 1969;105(7):1110–5.
[14] Morgan MA, Thurnau GR, Fishburne Jr JI. The fetal–pelvic index as an indica-
tor of fetal–pelvic disproportion: a preliminary report. Am J Obstet Gynecol
1986;155(3):608–13.
[15] Bland JM, Altman DG. Statistical methods for assessing agreement between two
methods of clinical measurement. Lancet 1986;1(8476):307–10.
[16] Sporri S, Thoeny HC, Raio L, Lachat R, Vock P, Schneider H. MR imaging pelvime-
try: a useful adjunct in the treatment of women at risk for dystocia? AJR
2002;179(1):137–44.
[17] 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(5 Pt 1):919–26.
[18] Pfammatter T, Marincek B, von Schulthess GK, Dudenhausen JW. MR pelvimetric
reference values. ROFO 1990;153(6):706–10.
[19] McKenzie CA, Lim D, Ransil BJ, et al. Shortening MR image acquisition time
for volumetric interpolated breath-hold examination with a recently devel-
oped parallel imaging reconstruction technique: clinical feasibility. Radiology
2004;230(2):589–94.