Levator ani thickness variations in symptomatic and asymptomatic ...

American Journal of Obstetrics and Gynecology (2004) 191, 856e61
IMAGING
Levator ani thickness variations in symptomatic and
asymptomatic women using magnetic resonance-based
3-dimensional color mapping
Lennox Hoyte, MD, a Marianna Jakab, MS, a Simon K. Warfield, PhD, a
Susan Shott, PhD, b George Flesh, MD, a Julia R. Fielding, MD c
Harvard Medical School, Boston, Mass, a Rush University, Chicago, Ill, b and University of North Carolina at Chapel
Hill, Chapel Hill, NC c
Received for publication December 28, 2003; revised March 31, 2004; accepted June 21, 2004
KEY WORDS
Levator ani
Magnetic resonancebased3dimensional
reconstruction
Color thickness
mapping
3-dimensional slicer
Pelvic organ prolapse
Pelvic floor
dysfunction
Magnetic resonance imaging (MRI) has greatly
improved our ability to study the organs and tissues of
the pelvis in women. Standard 2-dimensional MRI has
been used to assess the anatomy of the female pelvic
floor in cadavers and living women. 1-7
This research was supported by NIH P01 CA67165-06 and P41
RR13218, the June Allyson Foundation and the Society of Uro-
Radiology.
Reprints are not available from the authors.
0002-9378/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.ajog.2004.06.067
www.elsevier.com/locate/ajog
Objective: This study was undertaken to develop and test a 3-dimensional (3D) color thickness
mapping technique on levator ani imaged with magnetic resonance imaging (MRI).
Methods: Supine MRI datasets from 30 women were studied: 10 asymptomatic, 10 with
urodynamic stress incontinence, and 10 with pelvic organ prolapse. Levators were manually
outlined, and thickness mapping applied. Three-dimensional models were colored topographically,
reflecting levator thickness. Thickness and occurrences of absent levator substance (gaps)
were compared across the 3 groups, using nonparametric statistical tests.
Results: Color thickness mapping was successful in all subjects. There were statistically
significant differences in thickness and gap percentages among the 3 groups of women, with
thicker, bulkier levators in asymptomatic women, compared with women with prolapse or
urodynamic stress incontinence.
Conclusion: Color thickness mapping is feasible. It may be used to compare levators in
symptomatic and asymptomatic women, to study relationships between levator thickness and
pelvic floor dysfunction. This technique can be used in larger studies for hypothesis testing.
Ó 2004 Elsevier Inc. All rights reserved.
In seeking to better understand the specific anatomic
defects that may correlate to female pelvic floor dysfunction,
our group evaluated the morphology, volume, and
integrity of the levator ani and the bladder neck, using
reconstructed 3-dimensional (3D) MR-based models in
living women. Fielding et al 8 demonstrated the feasibility
of the 3D technique and yielded early estimates of the
normal range of levator volume (39 G 57 mL) in a group
of 10 asymptomatic women aged 22 to 33 years. Our
subsequent study used MR-based 3D reconstruction
techniques to evaluate a group of 30 women, divided

Hoyte et al 857
Figure 1 A reconstructed levator is shown in brown, with the
symphysis (S). Views of the front, left, and right sides are
shown. In the front view, the inner layer faces the label I, and
the outer layer faces the label O. Additional images and video
can be found online at www.us.elsevierhealth.com/ajog.
Figure 2 Three views of a color-mapped levator are shown.
The color bar on the left shows the thickness (in millimeters)
corresponding to the colors seen. Additional images and video
can be found online at www.us.elsevierhealth.com/ajog.
into 3 groups. Ten women were asymptomatic (ASY), 10
had urodynamic stress incontinence (USI), and the final
10 had symptomatic pelvic organ prolapse (POP). 9 That
pilot data showed significant differences in levator
volume, integrity, and shape across the range of ASY
to USI to POP, suggestive of a continuum of disease from
ASY to USI to POP. During that study, our observations
of the 3D reconstructed images suggested that levator
ani thickness may be a strong marker for pelvic muscle
health, but we believed that our efforts for objectively
quantifying this parameter were unsatisfactory. Therefore,
we sought to develop a technique for describing
levator muscle thickness that was visually intuitive,
mathematically consistent, and feasible for research
comparison across large groups of subjects. This article
presents the results of our thickness mapping technique,
applied to evaluate and compare the thicknesses of
levators from the 30 subjects in our prior study. 9
Material and methods
This is a secondary analysis of a previously presented
data set. 9 In that study, we evaluated 3 groups of women,
10 ASY, 10 USI, and 10 POP, ranging in age from 38 to
78 years. Mean age in the ASY group was 51 years, with
a range of 41 to 69 years. Subjects in the USI group were
slightly younger, with a mean age of 48 years (range 38-
56 years). The POP group was slightly older, with a mean
age of 57 years (range 42-78 years). Of the 10 subjects
in each group, there were 4, 6, and 6 postmenopausal
women in the ASY, USI, and POP groups, respectively.
Among postmenopausal subjects, hormone replacement
was used by 1 in the ASY group, 2 in the USI group, and
1 in the POP group. If ‘‘estrogen-exposed’’ subjects are
defined to include premenopausal and estrogen-using
postmenopausal women, there were 7 estrogen-exposed
women in the ASY group, 6 in the USI group, and 5 in
the POP group. Further details of subject selection, MR
scanning protocol, image segmentation, and 3D reconstruction
techniques are further described in our previous
article. 9 Our group had previously developed
a technique for characterizing and color-coding bladder
wall thickness from computed tomographic (CT) images
in an attempt to screen for bladder tumors. 10,11 We
modified this technique to measure and color code
levator ani thickness.
Preparation of the MR data for color
thickness mapping
Levator ani muscles on the source axial MR images
are manually segmented. The ‘‘inner’’ and ‘‘outer’’ surfaces
of the levator are identified on each slice of the
segmented data. The inner surface is defined as that
part of the levator that faces medially or superiorly in
situ, and the outer surface is defined as the part that
faces laterally or inferiorly. A 3D reconstruction is then
made from the segmented data. This is clarified in
Figure 1, where a reconstructed levator is shown with
the inner and outer surfaces labeled.
Color thickness mapping algorithm
An algorithm is applied to each point (Pinner) on the
levator inner surface, measuring its distance from all
points on the outer surface. The distance to the closest
point on the outer surface is taken as the thickness of the
levator at point Pinner. After the thickness is determined,
each point (or pixel) on the reconstructed image is
colored to correspond to its thickness. The range of
colors used goes in the order blue, green, yellow, orange,
and red, where blue is the thinnest and red is the
thickest. Each levator is then inspected, and its thickness
at each point determined from its color. An example of

858 Hoyte et al
a color-mapped levator is given in Figure 2, along with
a legend for mapping the colors back into distances in
millimeters.
Data collection and statistical analysis
Each levator is then divided into 4 quadrants, left
anterior (LA), left posterior (LP), right anterior (RA),
and right posterior (RP). The left-right distinction is
made with respect to a sagittal plane running through
the mid-symphysis and mid-coccyx. The anterior-posterior
distinction is made with respect to a plane parallel
to the puborectalis, taken midway between the symphysis
and the tip of the coccyx. Generally speaking, the
anterior divisions correspond to the puborectalis portion,
and the posterior divisions correspond to the
ileococcygeus portions. The maximum thickness, minimum
thickness, and presence or absence of gaps (ie,
absent levator substance) is measured for each quadrant
of each levator.
To determine the thickness and gap frequencies, each
quadrant of each color-mapped levator is inspected
onscreen. The maximum and minimum thickness in
each quadrant is determined from the range of colors
seen in the quadrant, and this information is recorded.
The median thickness and range is calculated for each
quadrant by study group.
The presence or absence of ‘‘gaps’’ in the levator
substance of each quadrant of each levator is determined
by inspection and is recorded as a ‘‘1’’ if a gap (ie,
absence of levator substance) is present and a ‘‘0’’ if no
gap is present in that quadrant. An example of a ‘‘gap’’
in levator substance is given in Figure 3, and labeled.
Gap frequency was computed and recorded for each
quadrant by group.
SPSS for Windows (version 10, SPSS Inc, Chicago,
Ill) was used for data management and statistical analysis.
Because the noncategorical measurements had statistically
non-normal distributions, the nonparametric
Kruskal-Wallis test and Mann-Whitney test were used to
compare the 3 groups. Fisher exact test and Fisher
extended exact test were performed to compare the
groups with respect to the percentage with gaps in each
quadrant. A .05 significance level was used for all statistical
tests. No 1-sided statistical tests were performed.
Results
Medians, ranges, and statistical comparisons of maximum
and minimum levator thickness for the test groups
are presented in Table I for each quadrant of levator ani,
across the groups. Statistical comparisons of gap frequencies
for each quadrant among the 3 groups are
presented in Table II. The right anterior levator was
significantly thicker among ASY women when com-
Figure 3 A view of gapping in the left side of a color-mapped
levator. The gap is the area of absent levator substance seen at
the point of the arrow. Additional images and video can be
found online at www.us.elsevierhealth.com/ajog.
pared with women with USI or POP. The LA levator
was significantly thicker in ASY women when compared
with women with POP. Minimal levator thickness (ie,
the thinnest part of levator ani) was greater in the RA
portion in ASY women when compared with women
with USI or POP. Minimal thickness was also greater in
the LA portions of levator in ASY women compared
with women with POP. In addition, there was a lower
incidence of ‘‘gaps’’ in the levator substance on the LA
levator in ASY women compared with women with USI
or POP. All these differences were statistically significant.
Comment
These pilot study data demonstrate the usefulness of the
color thickness mapping technique as a way to mathematically
characterize and communicate information on
levator muscle thickness in different portions of the
muscle. In this experiment, the technique permitted us to
easily visualize and quantify muscle thickness. On the
basis of this experience, the technique appears to be
suitable for reviewing and comparing large numbers of
levators.
The results of this study suggest that the anterior
portion of the levator (ie, puborectalis) is bulkier
bilaterally in ASY women compared with those with

Hoyte et al 859
Table I Maximal and minimal thickness parameters by levator quadrant in ASY vs USI and POP groups
ASY USI POP
Median Range
P value:
P value: P value:
(mm) (mm) Median Range ASY vs USI Median Range ASY vs POP USI vs POP
RA-max 10.01 8.6-11.35 7.33 5.33-10.01 .001 6.00 0.00-11.3 .006 NS
RA-min 4.66 2.00-7.33 3.33 0.66-3.33 .004 2.00 0.00-4.66 .013 NS
RP-max 8.00 2.00-11.33 6.00 3.33-8.66 NS 4.60 2.00-10.0 NS NS
RP-min 2.00 0.66-4.66 2.67 0.66-3.33 NS 1.33 0.66-3.33 NS NS
LA-max 11.35 10.01-11.35 11.33 7.33-11.35 NS 7.33 3.33-11.35 .003 .022
LA-min 4.66 2.00-7.33 3.33 2.00-4.66 .074 2.00 0.66-4.66 .007 .07
LP-max 7.33 4.66-11.35 8.66 6.00-11.35 NS 6.00 2.00-11.35 .059 .027
LP-min 2.00 0.66-4.66 2.00 0.66-3.33 NS 2.00 0.66-3.33 NS NS
RA-max, Maximal right anterior levator thickness; RA-min, minimal right anterior levator thickness; RP-max, min, right posterior levator maximal and
minimal thickness; LA-max, min, left anterior maximal and minimal thickness; LP-max, min, left posterior levator maximal and minimal thickness; NS, not
significant.
Table II Levator gap frequency by quadrant in ASY vs women
with USI and POP
ASY
%Gaps
USI
%Gaps
POP
%Gaps
ASY vs
USI
P value
ASY vs
POP
P value
RA 30 60 80 NS NS NS
RP 80 90 90 NS NS NS
LA 10 80 100 .005 .0005 NS
LP 90 90 100 NS NS NS
USI vs
POP
P value
POP, and bulkier on the right in ASY women
compared with women with USI. On the left, significant
increases in loss of levator substance are noted in
the anterior portions of the levators in women with
POP and USI, compared with ASY women. This
finding is also suggestive of decreased bulk in the
symptomatic groups. Possible explanations for these
findings include muscle atrophy caused by denervation
from childbirth injuries, or perhaps muscle wasting
caused by the loss of insertion points for the puborectalis,
also stemming possibly from childbirth injury,
as has been suggested by the results of Delancey et al. 12
It is also possible that these may be normal variations
in muscle parameters, as was seen by Tunn et al 7 when
they evaluated levator geometry in a group of nulliparous
women.
Because striated muscle atrophies with age, and
estrogen receptors have been detected in levator muscle
stroma and fascia, 13 it is reasonable to expect that
levator muscle bulk, and thus thickness, is likely to
decrease with age, and possibly also with estrogen
status. Age and hormonal status therefore represent
confounding factors in our analysis. There was considerable
age overlap among the groups, but the USI group
was the youngest, followed by the ASY and POP
groups, respectively. On the basis of these age distribu-
tions, it is possible that age-related muscle atrophy could
help to account for the differences in thickness and
‘‘gapping’’ between the ASY and POP (or USI and POP)
groups, but not between the ASY and USI groups.
Regarding hormonal status, there were 7 estrogenexposed
women in the ASY group, 6 in the USI group,
and 5 in the POP group. It is possible that hormonal
status might influence the comparison between the ASY
and POP (or USI and POP) groups, but less so the
comparison between ASY and USI groups. However,
these relationships are best explored by a larger welldesigned
study.
In either case, a reduction in puborectalis bulk is
noted in women with USI or POP, when compared with
ASY women. The clinical significance of this finding
remains to be clarified in larger studies.
In a multicenter study, Swift et al 14 found stage II
POP in 32% of women presenting to general gynecology
clinics. Their results were not generalizable to the
US population because of the racial and socioeconomic
composition of the study population. Still, it is probably
reasonable to expect a proportion of subjects with
stage II POP among our ASY group. If POP is
associated with a lower muscle bulk, then mean muscle
thickness in our ASY group would be pulled lower by
those ASY women with undiagnosed POP, and this
would tend to narrow the differences among the study
groups. If these ‘‘occult’’ POP subjects were excluded
from the ASY group, the overall thickness and bulk of
the levators in this group would be expected to increase,
strengthening the differences amon the groups.
This point reinforces the need for objective subject
characterization.
Regarding the gaps in the levator muscle substance,
the well-known chemical shift phenomenon can, by
artifact, ‘‘thin out’’ the muscle signal on MRI, depending
on the encoding direction of the MR scan, and the
spatial relationship of muscle to fatty tissue. For this

860 Hoyte et al
reason, one side (right) of the levator muscle tended to
appear thinner on the MRI than the other side (left).
Thickness comparisons between left and right sides were
not attempted because of this artifact. The artifactinduced
preferential thinning on the right should have
led to more ‘‘gaps’’ on the right, in places where the muscle
was truly thin to begin with. However, more gaps
were seen on the left, where the muscle was thicker.
This finding suggests that the ‘‘gaps’’ seen on the left
represent a true finding, unrelated to artifact. These gaps
may represent fibrosis or fatty infiltration of the muscle,
and remain to be elucidated.
The important topic of interobserver and intraobserver
measurement variability was addressed in the
original manuscript, covering the manual segmentation
used to calculate levator volume. 9 As noted there, the
interobserver bias was 1.17 mL, with limits of agreement
at G4.52 mL, and the intraobserver bias was
ÿ0.26 mL, and limits of agreement were G3.4 mL.
However, one could obtain identical volume measurements
from 2 different segmentation geometries. Consequently,
an appropriate reliability analysis needs to
consider the spatial overlap of the segmentations as well
as the actual numeric results (eg, volume or thickness).
That type of analysis was not available when we did
the current study, and is not included here. Since then,
a suitable reliability technique has been developed by
our colleagues at the Surgical Planning Laboratory, 15
and we look forward to testing it on our datasets.
The current 3D thickness analysis is more accurate
than a straight 2D analysis. This is true because a single
slice thickness measurement only considers adjacent
points within the plane of the slice, whereas the 3D
analysis also considers adjacent points outside of the
slice plane. Furthermore, variations in the MR slice
acquisition angle can introduce artifacts that alter the
thickness of the item under study. 16 These artifacts are
not present in the 3D thickness analysis. The 3D visual
renderings allow an observer to quickly localize and
compare thicknesses across large numbers of levators,
and this capability is not readily available in slice-byslice
thickness analysis. In addition, a full mathematical
description of the thickness at each point on the surface
is retained as part of the geometric description of the
color-mapped levator. Future algorithms can analyze
and compare such descriptions across groups of interest.
Our study has some limitations, and we plan to
remedy these in future work. First, our characterization
of our subjects was limited to urodynamic testing and
prolapse staging in the symptomatic groups only. In
future studies we plan to include a validated symptom
questionnaire, cough stress and urodynamic testing, and
POPQ staging in all of our study subjects. Finally, the
color-map analysis is limited by the speed of segmentation.
With the use of the manual technique, segmenta-
tions can proceed at the rate of under 1 person-hour per
dataset. This works out to about 2000 datasets per
person-year, assuming 40-hour work week. We are
currently evaluating methods for shortening the time
required to process each dataset. Despite these limitations,
our findings are potentially clinically relevant and
we plan to continue this work in larger studies in the
future.
Acknowledgments
We acknowledge the contributions of Dr Lore Schierlitz,
who helped to prepare the raw MR data for color
thickness analysis.
References
1. Strohbehn K, Ellis J, Strohbehn JA, Delancey JOL. Magnetic
resonance imaging of the levator ani with anatomic correlation.
Obstet Gynecol 1996;87:277-85.
2. Huddleston HT, Dunnihoo DT, Huddleston PM 3d, Meyers PC.
Magnetic resonance imaging of defects in DeLancey’s support
levels I, II, and III. Am J Obstet Gynecol 1995;172:1778-84.
3. Kirshner-Hermanns R, Wein B, Niehaus S, Schaeffer W, Jakse G.
The contribution of magnetic resonance imaging of the pelvic
floor to the understanding of urinary incontinence. Br J Urol
1993;72:715-8.
4. Healy JC, Halligan S, Reznek RH, Watson S, Phillips RKS,
Armstrong P. Patterns of prolapse in women with symptoms of
pelvic floor weakness: assessment of MR imaging. Radiol
1997;203:77-81.
5. Fielding JR, Versi E, Mulkern RV, Lerner MH, Griffiths DJ,
Jolesz FA. MR imaging of the female pelvic floor in the supine and
upright positions. J Magn Reson Imaging 1996;6:961-3.
6. Tunn R, Delancey JOL, Quint LE. Visibility of pelvic organ
support system structures in magnetic resonance images without
an endovaginal coil. Am J Obstet Gynecol 2001;184:1156-63.
7. Tunn R, Delancey JOL, Howard D, Ashton-Miller J, Quint LE.
Anatomic variations in the levator ani muscle, endopelvic fascia,
and urethra in nulliparas evaluated by magnetic resonance
imaging. Am J Obstet Gynecol 2003;188:116-21.
8. Fielding JR, Dumanli H, Schreyer A, Okuda S, Gering D, Zou K,
et al. MR-based three-dimensional modeling of the normal pelvic
floor in women: quantification of muscle mass. Am J Radiol
2000;174:657-60.
9. Hoyte L, Schierlitz L, Zou K, Flesh G, Fielding JR. Two- and 3dimensional
MRI comparison of levator ani structure, volume and
integrity in women with stress incontinence and prolapse. Am J
Obstet Gynecol 2001;185:11-9.
10. Jaume S, Ferrant M, Macq B, Hoyte L, Fielding JR, Schreyer A,
et al. Tumor detection in the bladder wall with a measurement
of abnormal thickness in CT scans. IEEE Trans Biomed Eng
2003;50:383-90.
11. Fielding JR, Hoyte L, Okon SA, Schreyer A, Lee J, Zou KH, et al.
Tumor detection by virtual cystoscopy with color mapping of
bladder wall thickness. J Urol 2002;167:559-62.
12. DeLancey JOL, Kearney R, Chou Q, Speights S, Binno S. The
appearance of levator ani muscle abnormalities in magnetic
resonance images after vaginal delivery. Obstet Gynecol
2003;101:46-53.
13. Copas P, Bukovsky A, Asbury B, Elder RF, Caudle MR.
Estrogen, progesterone, and androgen receptor expression in

Hoyte et al 861
levator ani muscle and fascia. J Womens Health (Larchmt)
2001;10:785-95.
14. Swift SE, Schaffer J, Bland D, O’Boyle A, Woodman P, Valley M,
et al. Pelvic organ support study (POSST): distribution of pelvic
organ support stages by POPQ in a general gynecologic population.
Pelvic Med Surg 2003;9:207.
15. Warfield SK, Zou KH, Wells WM. Simultaneous truth and performance
level estimation (STAPLE): an algorithm for the validation
of image segmentation. IEEE Trans Med Imag 2004;23:903-21.
16. Hoyte L, Ratiu P. Linear measurements in 2-dimensional pelvic
floor imaging: the impact of slice tilt angles on measurement
reproducibility. Am J Obstet Gynecol 2001;185:537-44.