adjustment was particularly important in the retroverted uterus, where the orientation of the cer- vix is anterior. The position of the coil relative to the cervix was.
607
The Uterine Cervix on In Vitro and In Vivo MR Images: A Study of Zonal Anatomy and Vascularity Enveloping Cervical Coil
N.
OBJECTIVE.
deSouza1
M.
I. C. Hawley2
receiver
J. E. Schwieso1
from
D. J. Gilderdale3
w. P. Soufter4
The purpose
omy and vascularity which
to interpret
SUBJECTS cytologically that
was
subtle
normal cervices placed intravaginally
ilarly studied,
RESULTS.
on in vivo and in vitro
the cervix. These appearances
AND METHODS.
images were obtained.
surrounding
of this study was to characterize
of the cervix
coil surrounding
Using
changes
in early
and imaging
and
enveloped
appearances
In the in vivo studies, the high-signal
the normal
MR images
provide
central
zonal anat-
obtained
a normal
with
a
data base
neoplasia.
Thirteen women of reproductive were imaged with a ring-design the cervix.
Seven uterine specimens
an
Ti-
age with clinically solenoid receiver and
T2-weighted
and coil axial
resected for benign disease were sim-
were correlated
with histologic findings. and two stromal zones identified. Unlike the uterine body, they
the endocervical canal
were
mucosa
could be differentiated on both Ti- and T2-weighted images, on which the inner ring had a low signal and the outer ring had an intermediate signal intensity. The outer zone was highly vascularized, with inflow effects from large vessels visible on single-slice scans.
enhanced
On administration
rapidly,
whereas
of gadopentetate
dimegiumine,
the outer stroma showed
the
endocervical
more gradual
mucosa
enhancement.
The
inner zone enhanced slowly relative to the outer zone. The parametrium was visualized up to 6 cm from the center of the coil, and adjacent colon, fat, and blood vessels were identified. Up to four lymph nodes less than i cm in diameter were seen in the parametrium of three subjects. In the in vitro studies, the endocervical mucosa was of high signal intensity. in the fibromuscular cervix, an inner low-signal ring correlated with a region of tightly packed stroma (fibroblasts and smooth muscle cells; cell count, 5900 ± 2376 nuclei/mm2) and the intermediate-signal-intensity outer zone corresponded to a
Received February 10, 1994; accepted after revision
April
19, 1994.
Presented of Magnetic 1 The
Royal
at the annual meeting of the Society Resonance,
Dallas,
March
Robert Steiner Magnetic Postgraduate
smith Hospital, United Kingdom.
Medical
Du Cane Address
2Department London 3Hirst
of
Resonance School,
Rd., London correspondence
Histopathology,
OHS, United
Research
Centre,
Royal
Post-
Hospital,
Kingdom. Elstree
Way, Boreham-
wood WD6 1RX, United Kingdom. 4
Department
of Obstetrics
and Gynaecology,
Royal Postgraduate Medical School, smith Hospital, London W12 OHS, United
0361-803X/94/1633-0607 © American Roentgen Ray Society
i994;163:607-612
Unit,
HammerW12 OHS, to N. M.
Medical School, Hammersmith W12
AJR
1994.
deSouza. graduate
region of more loosely packed stroma (cell count, 2199 ± 558 nuclei/mm2). Retention cysts were present in two multiparous cervices. CONCLUSION. These detailed appearances and enhancement patterns of the normal cervix need to be recognized so that subtle changes in locally invasive cervical neoplasia can be identified.
HammerKingdom.
MR imaging of the uterine cervix with a body coil delineates the low-signal inner fibromuscular stromal ring [1-6], but differentiation of the detailed zonal anchitectune seen on in vitro studies [7] is not possible. The use of an intravaginal coil that envelops the cervix improves the signal-to-noise ratio, which allows thin slices and small fields of view to be used [8]. However, the anatomic features of the cervix on images obtained with this technique have not been described, which may lead to difficulties in interpretation, particularly in subtle or early disease. The punpose of this study was to define the normal zonal anatomy and vasculanity of the cervix with an intravaginal coil in order to establish a baseline for the recognition of early invasive carcinoma. Healthy subjects and uterine surgical specimens resected for benign disease were imaged, and the MR appearances were comelated with histologic findings in the latter group.
DESOUZA
608
Subjects
used,
in this group
of subjects,
a design
with no angle
between
coil and handle was preferred. Further details concerning this coil have been described elsewhere [8]. Identical imaging parameters were used for in vivo and in vitro studies. All imaging was performed with a 1 .O-T unit with both Ti -weighted
80) spin-echo
AJR:163,
Dynamic
and Methods
A ring-design solenoid receiver coil mounted on a acetal homopolymen (Delrin) former (internal diameter, 37 mm) was placed around the cervix. The coil was attached to a 20-cm tubular delnn handle. Although coil designs with an angle of 45#{176} between the coil and handle have been
ET AL.
(660/20
[TRITE])
and T2-weighted
scanning
bolus injection gram
body
(SE) sequences.
weight.
mittee.
to carry
Thirteen
out
MR
imaging
of healthy
volunteers
had been obtained from the hospital’s healthy
women
21-40
years
for
ethics com-
old (mean,
30 years)
were imaged. All had had normal cervical smears within the previous 2.5 years (mean, 1.25 years). Eight were nulliparous and five were multiparous.
gradient-echo
and after
(GRE)
a
per kilo-
images
(50/15,
intensity
of the
T2-weighted
inner
and
outer
stromal
zones
on three
consecutive
images.
In Vitro Examinations Seven
surgical
uterine
specimens
obtained
from
premenopausal
women 33-48 years old (mean age, 43 years) who were undergoing for menorrhagia
also were
examined.
The specimens
were placed in normal saline and imaged 0.5-4.0 hr after resection. Methods of image analysis were identical to those used in the volun-
In Vivo Examinations
purposes
Single-slice
during
dimeglumine
900 flip angle, 10-sec acquisition time) were acquired in three slice positions over a period of 5 mm followed by a conventional transverse Ti-weighted SE MR image. Images were analyzed by measuring the width, area, and signal
hysterectomy
Permission
in two subjects
1994
(2500/
Axial 2.5-mm contiguous slices were obtained with a 192 x 256 matrix and two to four signal averages. A 10to 15-cm field of view (FOV) was optimal, as good parametnial detail was seen up to 6 cm from the center of the coil (12-cm FOV).
research
was done
of 0.1 mmol of gadopentetate
September
Nine were
in the follicular
and four in the secretory
phase
of the menstrual
phase. Two were taking combined
cycle
oral con-
teer group. The
cervix
versely
was
in 4.0-mm
The point of change
amputated,
placed
in formalin,
slices, and stained in the cellularity
sectioned
with hematoxylin
of the stroma
between
trans-
and eosin. the tightly
packed inner zone and the more loosely packed outer zone was not sharply defined, and its position was assessed by two observers in conference.
This
point
was then
used
to measure
the thickness
of
traceptives and one was taking progesterone only. Two had retroverted uteri; the uterus was anteverted in the others. Written informed
inner and outer zones with a vernier scale. The cellularity of the inner and outer zones was quantified by using an eyepiece and a gnid-cali-
consent was obtained
brated graticule smooth muscle
to count the number of stromal nuclei (fibroblasts and cells) per unit area. Care was taken to avoid areas
with
vessels
With
the patient’s
from all subjects. thighs
in abduction,
the coil was
inserted
into
the vagina and positioned around the cervix. A speculum was not required; digital insertion sufficed. The handle of the coil was immobilized in an external clamp stand (Fig. 1), with its base positioned under the subject’s thighs. (Earlier examinations without the clamp had produced considerable image degradation due to coil motion.) Coil insertion and removal caused only minor discomfort for the patient while the coil was passing through the introitus, but once the coil was in place, it was comfortable and well tolerated. No form of sedation was required, and termination of the examination was not requested by any subject. The position of the coil could be checked by digital intravaginal examination, and the angle and tilt in the vagina could be altered by raising, lowering, or tilting the point of attachment
to the
external
clamp.
This
adjustment
was
particularly
important in the retroverted uterus, where the orientation of the cervix is anterior. The position of the coil relative to the cervix was altered on inspection of the sagittal pilot images.
large
blood
Fig. 1.-Photograph angle,
of clamp stand for immobilizing and
tilt can
be adjusted
at both
intravaginal universal
coil joints
of inflammatory
cells.
Results In Vivo Examinations
Both Ti- and T2-weighted images showed the normal cenvix to consist of two distinct stromal zones and the mucosa surrounding the central canal. The mucosa was relatively high signal on both Ti- and T2-weighted images. Mucosal detail seen with this technique (pixel size, 0.6 mm2) showed a smooth and regular outline in the nullipanous cervix (Fig. 2) and a more irregular and indented outline in the parous cervix (Fig. 3). In addition, dilated glands filled with secretions were sometimes seen in the latter group. This degree of resolution is not provided by our standard body coil images (pixel size, 1 .3 mm), one of which is illustrated for comparison (Fig. 3C). The inner stmomal zone was best identified on T2-weighted images as a ring of low signal, whereas the outer ring was intermediate signal on Ti- and T2-weighted images (Figs. 2 and 3). Thickness and area measurements for the inner and outer zones obtained from the cervical coil images are listed in Table 1 On T2-weighted images, the contrast between inner and outer zones was 34.2 ± 12%. No obvious visible diffenences in the zonal anatomy were noted between women taking oral contraceptives and women who were not, or between the follicular and luteal phases of the menstrual cycle, although the small number of women studied make these comparisons difficult. The outer zone was highly vasculanized. Well-defined foci of slightly higher signal seen within the outer zone on T2weighted images were highlighted on single-slice fast scans because of inflow phenomena (Fig. 4A). On administration of 0.1 mmol of gadopentetate dimeglumine per kilogram body weight, brisk mucosal enhancement began at 30 sec and peaked at 120 sec after injection (Fig. 4). The fibromus.
In situ. Height, (arrows).
or groups
AJR:163,
September
IN VIVO
1994
Fig. 2.-Normal nulliparous cervix. A and B, Transverse Ti-weighted (660/20, and T2-weighted (2500/80, B) SE MR images
AND
IN VITRO
MR
OF
UTERINE
609
CERVIX
A)
obtamed with cervical coil show smooth mucosal outline of endocervical canal (small arrows) and low-signal inner stromal zone (large arrows).
C Fig. 3.-Normal
multiparous
A-C, Transverse outline
(small
arrows)
with both techniques
cervix.
Ti-weighted
(660/20, A) and T2-weighted (2500/80, B) SE MR images obtained with cervical coil show a more irregular mucosal than in Fig. 2. Transverse T2-weighted (2500/80) SE MR image (C) at same level obtained with body coil. Nabothian cyst is noted (large arrows), but cervical and parametrial detail is poorly seen in C in comparison with B.
TABLE
i : Width and Area Measurements
of Zones of Fibromuscular Inner Stroma
Measurements Obtained from:
Width
(cm)
Outer
Area
(cm2)
In vivo MR images
0.41 ± O.i4 (34 ± 9)
1 .60 ± 0.5 (23 ± 6)
In vitro
0.42
2.25
MR
Histologic
images
are means
± O.i i
(25 ± 9) 0.46 ± 0.83 (36±5)
specimens
Note-Values
Cervix
±
S D. Numbers
in parentheses
cular stroma enhanced more slowly, with the outer ring showing more prominent enhancement, making zonal differentiation maximum at 90 sec (Fig. 5). The parametnium was optimally visualized when a 12-cm FOV was used. Reduction in the peripheral signal did not make larger FOVs worthwhile. Adjacent rectosigmoid colon, obturator vessels, and bladder could be easily identified.
(29
±
0.97
±
i 0)
Width
(cm)
Stroma Area
(cm2)
0.79 ± 0.13 (66 ± 9)
5.42 ± i .0 (77 ± 6)
0.81
5.50
(66
0.83
±
0.16
± 9) ±
(7i
±
O.i6
± 10)
0.22
(64±5)
are percentage
Between diameter
s of total fibromuscular
cervix.
one and four lymph nodes were seen in three patients.
of less
than
1 cm
in
Vitro Examinations Two zones with central high signal representing mucosa and canal were identified on Ti- and T2-weighted images. In one
610
DESOUZA
ET AL.
AJR:163,
Fig.
4.-Dynamic
enhanced
normal cervix. A-Fi GRE (50/16,
September
MR
images
1994
of
9O flip angle) images ob-
tamed before(A)and at 30 sec (B), 60 sec (C), 90 sec (D), 120 sec (E), and 150 sec (F) after injectlon of gadopentetate dimeglumine (0.1 mmol/ kg body weight). Note strong mucosal enhance-
ment (arrows). G, T2-weighted
(2500/80)
image at same 1ev-
el for comparison.
surgical specimen, an endocervical polyp was visualized within the canal on the Ti-weighted images and confirmed on subsequent histologic examination. As on the in vivo images, the inner zone was of low signal and the outer zone was of higher signal; the contrast between the zones was 28.1 ± 14.5% (Fig. 6). Thickness and area measurements for the inner and outer stroma are given in Table 1 Multiple mucus-filled retention cysts were seen in the inner zone of two specimens. Histologically, the fibromuscular stroma of the cervix vanies in cellularity between the inner region immediately sun.
rounding the endocenvical mucosa and the outer more peripheral zone. In most instances, the transition from a tightly packed inner stromal pattern to a more loosely packed outer stroma (Fig. 7) is gradual and subtle. In two cases, the change was more abrupt. The innermost stroma has a significantly higher cell count (5900 ± 2376 nuclei/ mm2) than does the outer stroma (2199 ± 558 nuclei/mm2, p .002). It constitutes 36% of the total cervical width and, therefore, is likely to correspond to the low-signal-intensity inner zone seen on the MR images. Relatively high signal