Medical Ernest J. Wiesen, BSEE #{149}Jeffrey Graham G. Ashmead, MD .Alan
Improvement
R. Crass, MD M. Cohen, MD
Radiology
1991;
M. Bellon,
MD
#{149}
in CT Pelvimetry’
When computed tomographic (CT) digital radiography is used for pelvimetry, measurement error may occur. Geometric distortion in the latera! direction of the CT digital radiograph can lead to an error in any measurement of the transverse pelvic inlet. The authors measured the magnitude of this error on two scanners and present a general method for correction of this potential error. The authors also showed that an additional dose reduction is possible if the patient is imaged in the posteroanterior rather than anteroposterior projection. Index terms: Dosimetry 87.1211 #{149} Pelvis, measurement, Pregnancy, irradiation in
Errol
#{149}
#{149} Pelvis,
CT, 85.1281,
85.1211, 87.1281
.
W
have been using computed tomography (CT) to perform digital pelvimetry after the method of Federle et al (1), with use of two digital radiographs and one axial scan. The electronic cursor and the internal computer program supplied with the CT system are used to measure the distance. When clinical colleagues raised questions regarding the accuracy of the method, we analyzed a group of studies after the method of Ball and Marchbanks (2) and Beam et al (3). We found that although the sagittal measurements from the digital radiographs were very accurate, errors in the transverse measurements of over 10% were possible.
E
Our
clinical
staff
believes
that
this possible error is unacceptable. The inaccurate measurements were due to geometric distortion along the transverse plane of the CT digital madiograph. Along the direction of table travel (the longitudinal direc-
178:259-262
tion), very
the
x-ray
beam
is narrow,
with
small divergence. The longitudinal accuracy depends on the accuracy of measuring and displaying table position. However, the x-ray beam across the CT table (transverse direction) greatly diverges and causes geometric distortions. Most manufacturers
I
From
J.R.C., necology ter,
Case
the
Departments
E.M.B.,
A.M.C.)
(G.G.A.), Western
of Radiology
and Obstetrics
MetroHealth Reserve
Scranton Rd. Cleveland, March 16, 1990; revision vision received August Address reprint requests © RSNA, 1991
Physics
(E.J.W.,
and Gy-
Medical
University.
OH 44109. requested 1; accepted to E.J.W.
Cen3395
Received April 26; reAugust 7.
of CT
scanners
cali-
brate the system to make transverse measurements accurate in the plane perpendicular to the x-ray beam and passing through the CT center of rotation (COR). Therefore, any measurement made in this plane is very accurate in every direction. Any transverse measurement made in any other plane above or below the COR in the direction of the x-ray beam divergence will be in error. The amount and direction of the error depend on the distance and direction of the plane from the COR. Federle et al (1) reduced the error, a priori, by positioning the patient within the CT gantry so that the
transverse
measurement
points
were
at the COR. Adam et al (4) suggested a corrective method based on their experience. Their method references measurements to the table top and requires either a fixed standard table height for scanning or some comrection
for
table
height.
Most
currently
available CT scanners can perform digital radiographic imaging at vanous table levels, and the table level used cannot be ascertained from the resulting image, a factor resulting in possible
unacceptable
errors.
The
goals of the present study were to develop a generally applicable solution to improve accuracy in CT pelvimetry independent of manufacturer, table position, or patient habitus. Our method references all measurements to the COR plane of the scanner, the location with the least error. Previous reports (1,4-6) have shown that fetal doses are low in this examination. A recent review showed that it was acceptable to perform digital radiography in the anteropostemior (AP) projection. Because the fetus is anterior in the matemnal abdomen, potential fetal dose is maximized.
We
thought
that
sig-
nificant possible scanning,
dose reduction would be with posteroanterior (PA) as well as lowering the xray exposure. The x-ray exposure can be reduced in both of the digital madiographs and the axial scan until the images become too noisy for accurate identification of the anatomic landmarks necessary to make the measurements.
In this
study,
both
the
lowering of fetal dose with use of PA digital radiography and correcting inaccuracies in the transverse measumement were considered.
Abbreviations: center of rotation,
AP = anteroposterior, PA posteroanterior.
COR
=
259
MATERIALS
AND
Geometric
were
made
METHODS
distortion
measurements
by scanning
a wooden
phan-
tom (Fig la) in which lead beads cated 20 cm apart in the transverse tion
at the
COR
and
were lodirec-
at distances
of
10 cm
and 5 cm above and below the plane the COR. For accuracy measurements the direction direction),
lead
of table beads
travel were
of in
(longitudinal located 10 cm
apart when projected onto the COR plane at varying distances from the COR. A more detailed drawing of the phantom is shown in Figure lb. A 3/4-inch plywood board was perpendicular to and angled 25#{176} relative
to the
center
line
of the
b.
a.
hori-
zontal 3/4-inch plywood board. Lead beads were placed on the slanted vertical board. The horizontal board of the phantom was placed in the exact center of the CT scanning
table.
The
center
of the
aligned with the longitudinal of the phantom. The table until
the
center
the COR scan was was used
was
center line was elevated
phantom
was
in
of the scanner. A lateral digital obtained. The electronic cursor to measure distance (Fig lc, id).
Two scanners nique: tion,
of the
table
were
tested
the Picker 1200 Highland Heights,
SX
with
this tech-
(Picker IntemnaOhio) and GE
9800 (GE Medical Systems, Milwaukee). On the basis of these measurements, curves can be obtained and correction tors
error fac-
calculated.
To determine simply
correction
divided
each
factors,
of the
five
plot could
by the at each
measure-
from
the
ners
COR.
tested
Dose use
magnification factor was distance. We chose to
a curve so that the be easily obtained
The
are
correction at any
curves
shown
in
measurements
of a female
Research,
Rando
and fluoride
simeter
chips
with
all
patient
(Alderson
The
phantom water-filled similar
was
inbags. to that used we
measurements,
the
measured
provide a constant tube was located phantom
was
doses
scanned
tion for the PA view AP view. Normally, with the tube under patient
also
distance, under the
used docali-
in the and prone pelvimetry the patient
supine
To
the posi-
for the is done and the
RESULTS scanners tested showed a in accuracy of measurement
of approximately timeter With
260
the
above or below patient supine,
Radiology
#{149}
1.5%
error
per
the COR. the mea-
depending
cen-
too
large
in the
AP
(OR
or
on whether
is scanned
the
or PA
of magnification factors found
the error is approximately +9%. measurement can be multiplied correction
factor
(.91) and the error pected considerable
supine.
Both variation
either
for
this
reduced. variation
small
errors
with
()
/
This by
Curve (t%t,
distance
-10
2
.5
change
the
through the tal radiograph
The
center of the represents
2
-4
()
2
4
6
10
8
of the of the
scans.
-6
Figure 2. To successfully use curves, one must pay attention to the position of the xray tube, the COR, and the measured point to be corrected.
scanners. The measurement that should be corrected is shown in Figure 3. The localization of the transverse inlet on the lateral scan is via the methods of Bean et al (3). The table height and the patient position should not
two
between
‘.-Corettu.n Fator
We exin the
either
Magn.t.(atu)n
are
geometric error between the two scanners tested. However, the variation was small, and we used Figure to find the correction factor, with
very
1.14
1.
shown in Figure 2. If the transverse diameter of the pelvic inlet is 6 cm closer to the x-ray tube than the COR,
the
change.
the x-ray table, and
be
plane. The degree and the correction
maxi-
brated with a 660 system (Victoreen, Cleveland). If the distance changes between the x-ray source and the patient’s skin,
may
too small,
2.
Shearer (5), except thermoluminescent for
surement
scan-
made
Conn).
mum thickness of the creased to 23 cm with The method used was by Moore lithium
Figure
phantom
Stamford,
surement.
factor distance
for the
were
d.
Figure 1. (a) Photograph of the wood phantom used for the test. (b) Detailed drawing shows the location of all lead beads. Beads in the central row are 10 cm apart when projected on the plane of the COR. All dimensions are in centimeters. (c) Lateral CT scan of the phantorn. The outside lead beads are 20 cm apart. The effect of geometric distortion is shown by the varying distance between beads. Lines 1, 2, and 3 are the actual measurements, with the measured distance shown at lower right. (d) Same image as c, except measurement is in direction of table travel. All beads in that direction are equally spaced. Line I is an actual mea-
we
ments by the measurement at the COR (20 cm). This would yield the magnification at -10, -5, 0, +5, and +10 cm from the COR. To obtain the correction factors,
1 divided corrected
c.
line lateral digithe plane
graph.
COR
for
and
the
y coordinates
COR is represented of the y coordinate between the COR inlet of
measurement
lateral the
known
PA
CT scanners
Our
digital
radio-
display
the
x
the cursor. The when the value is 0. The distance and the transverse of
can
be
located
on
scan
(Fig 4a). The location tube and detector must be to determine which correc-
January
1991
0.
3.
(a) PA digital
radiograph
C.
used
to measure the transverse the measured distance shown at lower right. (b) Lateral view shows table, this measurement should not need correction. Diagonal white shows measured distance between ischial spines. Horizontal white Figure
inlet diameter. Horizontal white line is the actual measurement, with measurement of AP diameter. If the patient is accurately centered on line shows actual measurement. (c) Axial section obtained at 40 mAs line shows actual measurement.
the
COR
. - .
#
1 1
-P,
b. Figure
,
4. (a) Diagram shows method for locating the level of the transverse inlet diameter. The COR is located by drawing a horizontal line (dashed line) at the center of the lateral digital radiograph. The AP inlet (line 1) is drawn. Line 2 is drawn at the cord of the sacrum. Line 3 is constructed, parallel to line 2, to pass through the ischial spine. The intersection of lines 1 and 3 (point A) is the level of the transverse inlet diameter. The perpendicular distance between the COR and point A is the distance (d) to be used to find the correction factor in Figure 2 and corrects the measurement made when the image in Figure 3 is used. (b) Measurements as made on the CT scanner display. Lines are the same as those in a. The only distance displayed at lower right (28 mm) is the distance between point A and the COR.
Figure
5.
Five lithium fluoride dosimeter chips were location (1-12) for both AP
each aging.
tion
factor
must
that
dardized,
with
the
supine To locate
verse
and
inlet
draws
lateral
be applied.
the the
x-ray
nec-
be stantube
patient. the level
of the
diameter,
one
a line
digital
We
procedure
under
at the
first
COR
radiograph.
locates
on the Line
1 (Fig
3 is constructed tangent to the ischial spine and parallel to line 2. The intersection of lines 1 and 3 (point A) is the level of the transverse inlet diam-
Volume
The
perpendicular 178
Number
#{149}
the
distance 1
be-
COR
the distance rection factor
correct trans-
4a) is drawn at the AP inlet. Line 2 is drawn at the cord of the sacrum. Line
eter.
tween
the
the image 4b shows
The
a connection
measurement in Figure a clinical
distance
inlet and 2, 2.8 cm
line and point A is (d) used to find the conin Figure 2 that will
between
made 3a is used. measurement.
the
transverse
the COR is 2.8 cm. toward the detectors
factor
when Figure
In Figure yields
of 1 .04. Therefore,
the transverse measurement is multiplied by 1.04. Our measured doses did not deviate significantly from those obtained in previous work (5,6). The location
within the phanin Figure 5. The methe total dose from and either the PA or
of the dosimeters tom are shown suits
represent
both
the
lateral
AP digital
ommend
thermoluplaced in and PA im-
minescent
radiograph
but
not
the
axi-
al scan (Table). was used. Moore
Only and
one axial Shearer
scan (5) me-
ported
fetus
was
enough
that
the
low
to be included in the axial scan in only 55% of these examinations. We did not include the axial scan because only a single section of the fetus was
exposed in
only
ceives whole-body
to direct 55%
most
radiation
of the
cases.
of the
radiation
digital
and The
then
fetus
from
me-
the
scanning.
It is difficult to find the exact dose difference between AP and PA views because the exact fetal location within the abdomen is not known. If the dosimeter locations 3 and 5 in Figure
Radiology
261
#{149}
5 are
assumed
to more
accurately
rep-
resent the location of the fetus, the fetus receives from one-third to onehalf the radiation dose in PA digital radiography. If the average of the eight dosimeter locations is used, the dose is about half. Peak doses also showed about a 50% reduction. Briefly,
the
fetus
would
receive
approxi-
mately half the radiation dose if PA digital radiography were used instead of AP, if all other positioning conditions are kept constant. DISCUSSION To find the minimum acceptable exposure for the axial scan, a female pelvic phantom was scanned several times, starting at 240 mM and at decreasing milliampere-second values until it was thought that the image was usable despite the noise. Although a 30-mAs setting is generally used for digital scanning, the technologist will increase this setting to yield an acceptable image if the patient is unusually thick. A typical low-dose axial scan is shown in Figure 3c. Although any table height can be used, it is important to remember that neither the patient nor the pa-
262
Radiology
#{149}
tient table can be moved between the lateral and PA digital radiographic studies. The scans must be obtained at right angles to each other. When done properly, there is a horizontal line on the lateral scan that represents the plane of the COR on the PA scan. On most CT scanners, this will be a horizontal line in the center of the lateral image. It is possible that the CT system may be out of calibration on that some manufacturers shift the image. The COR should be yenfied by obtaining a digital radiographic scan of a wire passing through the COR. The wine should appear in the center of the image on both the PA and lateral digital radiographs. The correction factors for CT scanners made by other manufacturers should be determined. Possible modification of the geometry of the tested CT scanners indicate that it would be safer if correction factors were detemmined for every scanner used for digital pelvimetry. The correction and magnification factors can be determined as described previously. The technique we used combines the techniques of Federle et al (1) and Beam et al (3), except that we reference all measurements to the COR instead of the table top. Use of this technique will increase the accuracy
of the measurement of the transverse diameter of the pelvic inlet. The accuracy of the AP diameter of the pelvic inlet is already good if the patient is centered properly on the table. An additional reduction in fetal dose is possible if PA, instead of AP, digital radiography is used. N Acknowledgments: technologists MetroHealth tance.
The authors of the Department Medical Center,
thank
the CT
of Radiology. for their assis-
References 1.
2.
3.
4.
5.
Federle MP, Cohen HA, Rosenwein MF, Brant-Zawadski MN, Cann CE. Pelvimetry by digital radiography: a low-dose examination. Radiology 1982; 143:733-735. Ball RP, Marchbanks 55. Roentgen pelvimetry and fetal cephalometry: a new technic. Radiology 1935; 24:77-84. Bean WJ, Cook RT, Eavenson LW, Bristow U. Pelvimetry. Radiol Clin North Am 1967; 5:29-46.
Adam Y, Alberge S. Castellano M, Kassab M, Escude B. Pelvimetry by digital radiography. Moore mates
Clin Radiol 1985; 36:327-330. MM, Shearer DR. Fetal dose for CT pelvimetry. Radiology
esti1989;
171:265-267.
6.
Claussen C, Kohler D, Christ F, Golde G, Lockner B. Pelvimetry by digital radiography and its dosimetry. J Perinat Med 1985; 13:287-291.
January
1991