Original Article

Value of Magnetic Resonance Imaging in Diagnosis of Adenomyosis

and Myomas of the Uterus
Charalampos P. Stamatopoulos, MD, PhD, Themistoklis Mikos, MD, PhD*,
Grigoris F. Grimbizis, MD, PhD, Athanasios S. Dimitriadis, MD, PhD,
Ioannis Efstratiou, MD, PhD, Panagiotis Stamatopoulos, MD, PhD,
and Basil C. Tarlatzis, MD, PhD
From the First Department of Obstetrics and Gynecology, Papageorghiou General Hospital (Drs. C.P. Stamatopoulos, Mikos, Grimbizis, P. Stamatopoulos,
and Tarlatzis), Department of Radiology, AHEPA Hospital (Dr. Dimitriadis), Aristotle University of Thessaloniki, and Department of Pathology,
Papageorghiou General Hospital (Dr. Efstratiou), Thessaloniki, Greece.

ABSTRACT Study Objective: To estimate the diagnostic performance of magnetic resonance imaging (MRI) in detection of myomas and
adenomyosis of the uterus.
Design: Prospective cohort observational study (Canadian Task Force classification II-2).
Setting: Department of obstetrics and gynecology, tertiary academic hospital.
Patients: One hundred fifty-three consecutive women with an enlarged uterus accompanied by gynecologic symptoms and/or
with an asymptomatic pelvic mass.
Intervention: Total abdominal hysterectomy. All patients underwent MRI before the operation.
Measurements and Main Results: The sensitivity, specificity, positive, and negative predictive value of MRI for the diag-
nosis of uterine pathology was calculated using histologic findings as the standard criterion for final diagnosis. Receiver op-
erating characteristics curves were constructed to describe the diagnostic performance of MRI. In the diagnosis of myomas,
MRI demonstrated sensitivity of 94.1%, specificity of 68.7%, PPV of 95.7%, and NPV of 61.1%. In the diagnosis of adeno-
myosis, MRI demonstrated sensitivity of 46.1%, specificity of 99.1%, PPV of 92.3%, and NPV of 88.5%. The area under the
curve (AUC) for the diagnostic performance of MRI in the detection of myomas and adenomyosis was 0.81 and 0.73, respec-
tively. Uterine sarcoma was diagnosed in 5 patients; in these cases, MRI demonstrated sensitivity of 60.0%, specificity of
99.2%, PPV of 75.0%, and NPV of 98.4%. The AUC for MRI in the diagnosis of uterine sarcomas was 0.80.
Conclusions: MRI exhibits a high AUC for the diagnosis of both adenomyosis and myomas. The PPVof MRI in the diagnosis
of adenomyosis and myomas of the uterus is high as well. MRI seems to be a useful technique in everyday clinical practice in
the diagnostic approach of these common conditions, enabling clinicians to select the most appropriate management. Journal
of Minimally Invasive Gynecology (2012) 19, 620–626 Ó 2012 AAGL. All rights reserved.
Keywords: Adenomyosis; Magnetic resonance imaging; MRI; Myoma; ROC analysis; Uterine tumors
Use your Smartphone
to scan this QR code
DISCUSS You can discuss this article with its authors and with other AAGL members at and connect to the
http://www.AAGL.org/jmig-19-5-12-00088 discussion forum for
this article now*
* Download a free QR Code scanner by searching for ‘‘QR
scanner’’ in your smartphone’s app store or app marketplace.

Uterine myomas and adenomyosis are 2 of the most

The authors have no commercial, proprietary, or financial interest in the
products or companies described in this article.
Corresponding author: Themistoklis Mikos, MD, PhD, First Department of
Obstetrics and Gynecology, Papageorghiou General Hospital, Ring Road,
Nea Efkarpia, Thessaloniki 56403, Greece.
E-mail: themis.mikos@gmail.com
Submitted February 12, 2012. Accepted for publication June 21, 2012.
Available at www.sciencedirect.com and www.jmig.org
1553-4650/$ - see front matter Ó 2012 AAGL. All rights reserved.
http://dx.doi.org/10.1016/j.jmig.2012.06.003
common benign uterine conditions that can cause morbidity
in premenopausal women. Conservative management failure
or bothersome symptoms mandate surgical management of
myomas in a large percentage of this population. The inci-
dence of myomas can be as high as 60%, and race/ethnicity,
heredity, nulliparity, obesity, polycystic ovarian syndrome,
diabetes, and hypertension seem to be associated with
Stamatopoulos et al. MRI in Adenomyosis and Uterine Myomas
increased risk of myomas [1–4]. Routine detection of
myomas is achieved at ultrasonography (US). Although
US has multiple advantages insofar as cost, patient-
friendliness, availability, no radiation, repeatability, and
learning curve, in certain circumstances, it falls short in its
ability to map multiple myomas or lesions in large uteri [5].
However, contemporary clinical practice demands precise
diagnosis so that patients receive the most appropriate and
individualized treatment [6–8].
Adenomyosis is a unique condition that puzzles both
patients and health care providers. Adenomyosis can either
be diffused or localized, in the form of adenomyoma [9].
The incidence of adenomyosis is estimated to be 25% to
35% in women undergoing hysterectomy because of benign
gynecologic conditions, although in some studies, it has
been reported to be as high as 70%. The condition is difficult
to diagnose because it demands biopsy, which in most cases is
undesirable and/or difficult to perform, and treatment is empir-
ical and not conclusive unless it involves removal of the uterus
[10]. Ultrasonography, hysteroscopy, and laparoscopy facili-
tating myometrial biopsy, and computed tomography have
been proposed as routine tests for diagnosis of adenomyosis
[11]. However, none of these diagnostic methods has high sen-
sitivity to enable correct identification of adenomyosis.
Magnetic resonance imaging MRI has been used in the
past 3 decades for accurate management of soft tissue dis-
eases. The MRI diagnosis of myomas is consistently high,
and seems to be superior to US in prospective comparative
studies [12,13]. Similarly, compared with US, when
adenomyosis is suspected, MRI enables more accurate
diagnosis of the disease [14]. There is no doubt that prospec-
tive comparison of MRI vs hysterectomy specimens (crite-
rion standard) seems to be the most appropriate method of
determining the diagnostic accuracy and clinical perfor-
mance of MRI in the diagnosis of myomas and adenomyosis.
The objective of the present study was to estimate the di-
agnostic performance of MRI in detection of myomas and
adenomyosis of the uterus using receiving operating charac-
teristics (ROC) curves.
Materials and Methods
Study Design and Power Analysis
This prospective cohort study was designed and executed
in the Department of Obstetrics and Gynecology of a tertiary
academic hospital. Ethics committee approval was obtained.
The ROC curve is a graph that plots the sensitivity of a diag-
nostic test as a function of its corresponding specificity
(1-specificity or false-positive rate). In tests that are both sen-
sitive and specific, the area under the ROC curve (AUC)
approaches 1.0, whereas in tests neither sensitive nor spe-
cific, the AUC lies near the 45-degree diagonal line, which
represents poor performance or performance not better than
that could be expected by chance. Using ROC analysis, re-
searchers can indicate a diagnostic test with good positive
predictive value (PPV) (i.e., the possibility of a lesion found
621
at MRI to be correctly diagnosed) and, therefore, use
resources in the most cost-effective way [15,16].
To calculate the necessary sample size, a preliminary
power analysis was performed on the basis of published
literature describing the accuracy of MRI in the diagnosis
of myomas [17–20] and adenomyosis [19,21–24].
The present study was performed to demonstrate that
MRI produces significantly different AUCs (alternative
hypothesis) in the diagnosis of either myomas or adenomyo-
sis when compared with the null hypothesis value 0.5. The
AUC from similar studies in which MRI was used for diag-
nosis of myomas was 0.92 [17–19]; to achieve an AUC
.0.50 in the present study for the diagnosis of myomas,
more than 40 patients needed to be recruited (type
I error 5 0.05, two-sided; type II error 5 0.05). Similarly,
the AUC from similar studies in which MRI was used for
diagnosis of adenomyosis was 0.81 [17,19–22]; to achieve
an AUC .0.50 in the present study for diagnosis of
adenomyosis, more than 80 patients needed to be recruited
(type I error 5 0.05, two-sided; type II error 5 0.05).
Thus, the minimum population needed to reach statistical
significance was 81 patients.
Patients
Women with heavy menstrual bleeding and/or a pelvic
mass were assessed in the outpatient gynecology depart-
ment. Consecutive patients were recruited, and a conve-
nience sample (simple random sample) was assigned to
undergo further scheduled interventions.
Inclusion criteria were bulky uterus (larger than 10
weeks’ gestational size) with no history of previous histo-
logic investigation; no desire for further childbearing; and
consent for total abdominal hysterectomy to establish the
histologic diagnosis.
Exclusion criteria included pregnant status; desire to re-
tain fertility; preoperative diagnosis of malignant disease
requiring hysterectomy; history of minimally invasive treat-
ment of menorrhagia (e.g., endometrial ablation or endome-
trial resection); need for uterine morcellation during surgery,
which hampers standard histopathologic examination; and
medical contraindications to surgery.
Signed informed consent was obtained from all patients.
All patients gave a detailed medical and gynecologic history,
and underwent pelvic examination and transvaginal US.
Magnetic resonance imaging, surgery, and histopathologic
analysis were performed independently and without knowl-
edge of the other investigators’ diagnosis. When the preop-
erative investigations were completed, the patients gave
consent for total hysterectomy and further histopathologic
examination of the uterine specimens.
Clinicians
All hysterectomy specimens were examined by a single
pathologist (I.E.), all MRI images were evaluated by a single
622
Table 1
Correlation of MRI findings with histopathologic findings
Histopathologic findings
MRI findings Myoma only Myoma 1 adenomyosis
Myoma only 89 14
Myoma 1 adenomyosis 1 8
Adenomyosis only 0 2
Cancer 1 0
Other 4 0
Total 95 24
MRI 5 magnetic resonance imaging.
Bold indicates correctly identified cases.
radiologist (A.S.D.) with special interest in pelvic MRI and
extended experience in the diagnosis of myomas and adeno-
myosis, and all operations were performed in a standard
fashion by 2 gynecologists (G.F.G., P.S.).
Magnetic Resonance Imaging
All patients underwent MRI of the lower abdomen and
pelvis (Magnetom Impact 1.0 T; Siemens AG, Erlangen,
Germany). During MRI, the following sequences were
obtained: T1-weighted image, to indicate the borders and
the anatomy of the internal organs; T2-weighted image, to
differentiate the layers of the uterus and enable diagnosis
of any abnormalities; sections after intravenous administra-
tion of gadolinium–diethylene triamine pentaacetic acid, to
differentiate the layers of the uterus and to stage neoplastic
disease, if found. During MRI, longitudinal, axial, and obli-
que sections were routinely obtained. Each examination was
completed within 30 to 45 minutes.
For diagnosis of adenomyosis, the junctional zone con-
tours were reported as uniform or not uniform insofar as
thickness. The thickness was measured at the anterior and
posterior walls at the thickest part of the zone. When the
thickness was .12 mm, focal not well-demarcated areas
were present in the myometrium, and the junctional zone
was nonuniform, adenomyosis was diagnosed [25].
For diagnosis of leiomyomas, disturbance of the normal
uterine contour was used. Leiomyomas typically produce
distinct low-signal intensity with a speckled appearance on
T2-weighted images that exhibits a ‘‘pushing’’ border and
a rounded appearance. Diffuse uterine enlargement with
multiple nodules was also classified as leiomyoma. Presence
of edema, hyaline, cystic, and red degeneration assisted in
diagnosis of leiomyoma [26].
For diagnosis of uterine sarcomas, MRI findings of large
uterine masses with extensive hemorrhage and necrosis was
used. Massive uterine enlargement with irregular central zones
of low attenuation suggesting necrosis and hemorrhage indi-
cated leiomyosarcoma, although these findings are not specific
for sarcoma and may be present with myoma degeneration,
which is more frequent. When extensive invasion into the my-
Journal of Minimally Invasive Gynecology, Vol 19, No 5, September/October 2012
Adenomyosis only Cancer Other Total
0 1 3 107
1 0 0 10
1 0 0 3
0 3 0 4
0 1 6 11
2 5 9 135
ometrium was found, low-grade endometrial stromal sarcoma
was suspected. When infiltration of the myometrium was in
a more destructive manner, a voluminous polypoid mass was
expanding the endometrial cavity, continuous extension into
the adjacent structures and heterogeneous signal intensity on
both T1- and T2-weighted images was found, high-grade en-
dometrial stromal sarcoma was suspected [27].
Surgery
Four-pedicle total abdominal hysterectomy with or with-
out salpingo-oophorectomy was performed in a standard
fashion. Uterine manipulations such as morcellation, grasp-
ing of the organ with ‘‘destructive’’ surgical instruments,
electrosurgery, and high-energy ultrasonic diathermy were
not allowed; therefore, when their use was necessary, the
patient was excluded from the study.
Pathologic Analysis
Specimens were evaluated immediately after their re-
moval, before fixation, and their volume and weight were de-
termined. The specimens were cut in the mid-sagittal plane
into 7- to 12-mm sections. The diagnosis of myoma was
reached conventionally. The histologic diagnosis of adeno-
myosis was made when ectopic endometrium was recog-
nized R2 mm deep in the myometrium or .1 microscopic
field at 10-fold magnification from the endomyometrial
junction; adenomyotic foci were circumferentially sur-
rounded by bundles of hypertrophic smooth muscle cells at
hematoxylin-eosin staining; or stromal fibroblasts clearly
differed cytologically from the adjacent smooth muscle cells
[28]. When endometrial glands or stroma were diffusely dis-
tributed in the myometrium, diffuse adenomyosis was diag-
nosed. When circumscribed nodular aggregates of glands or
stroma were found within the myometrium, focal adeno-
myosis or adenomyoma was diagnosed. For diagnosis of ad-
enomyomas, cases were identified when a circumscribed
mass was found, composed of more than rare glands, pre-
dominantly of the endometrial type, and a stromal compo-
nent that consisted primarily of smooth muscle [29].
Stamatopoulos et al. MRI in Adenomyosis and Uterine Myomas 623

Table 2
Characteristics of 5 patients with uterine sarcoma

Ultrasonographic Magnetic resonance Histopathologic

Age, yr Parity Symptoms findings imaging findings findings
45 2 Menorrhagia Uniform enlargement Sarcoma Stromal sarcoma
58 1 Postmenopausal bleeding Uniform enlargement Sarcoma Stromal sarcoma
51 1 Incidental Myomas Myoma Stromal sarcoma
79 0 Postmenopausal bleeding Uniform enlargement Endometrial carcinoma Stromal sarcoma
57 1 Pelvic mass Myomas Sarcoma Leiomyosarcoma

Statistical Analysis Results

Data collection and initial statistical analyses were per-
formed using commercially available software: MS Office
(Microsoft Corp., Redmond, WA), SPSS version 17.0
(SPSS, Inc., Chicago, IL), and MedCalc version 9.5.2.0
(MedCalc Software bvba, Mariakerke, Belgium). Normal-
ity testing (D’Agostino and Pearson test) was performed
to determine whether data were sampled from a Gaussian
normal distribution. For purposes of the study, 2 ! 2 tables
were created whereby MRI diagnoses were compared with
histologic diagnoses and the Cramer V indicator (a decimal
ranging from 0 to 1 that measures the association between
2 variables; in the present study, the Cramer V was used to
measure the degree of relationship between MRI diagnosis
and histologic results) and the Cohen k, a statistical mea-
sure of the agreement of qualitative variables, which is
thought to be more robust because it takes into account
agreement occurring by chance, were used to evaluate
agreement between them. Sensitivity, specificity, and PPV
and negative predictive value (NPV) were calculated for
MRI using the histologic results as the criterion standard.
The ROC analysis was performed last, and ROC curves
for the diagnostic performance of MRI in detection of my-
omas, adenomyosis, and coexistence of myomas and ad-
enomyosis were constructed. Statistical significance was
considered at p , .05, and the 2-sided z-test was used to
evaluate the AUCs.
Of 153 women with gynecologic symptoms and/or an en-
larged uterus, 135 fulfilled inclusion and exclusion criteria
and agreed to participate in the study. The mean (SD; 95%
CI) age of the study population was 46.7 (11.20; 44.83–
48.65) years. Parity was 1.6 (1.04; 1.44–1.81) children (2
[0–5]). The referring symptom was heavy menstrual bleed-
ing in 78 patients, postmenopausal bleeding in 12,
abdominal heaviness, bloating, and suprapubic pain in
24, and pelvic mass in 9. In 5 patients, enlarged uterus
(.12-weeks gestational size) was an incidental finding.
At histopathologic analysis, myomas were diagnosed in
95 patients, adenomyosis in 2, coexistent myomas and
adenomyosis in 24, and uterine sarcoma in 5 (endometrial
stromal sarcoma in 4 and leiomyosarcoma in 1). In 9 patients
other diagnoses were made, including ovarian disease in 4,
uterine cervical disease in 2, endometrial disease in 2, and
stromal disease in 1. Thus, the incidence of myoma was
88.15% (119 of 135; 95% confidence interval [CI], 81.47–
93.07), of adenomyosis was 19.26% (26 of 135; 95% CI,
12.99–26.93), and of uterine sarcoma was 3.70% (5 of 135;
95% CI, 1.23–8.44). At MRI, myomas were found in 117 pa-
tients, adenomyosis in 13, and uterine sarcoma in 4, whereas
in 11, another diagnosis was made. Detailed results of MRI
and histopathologic analysis are given in a double-input table
(Table 1), and the clinical, radiologic, and histopathologic
findings of the 5 uterine sarcomas are given in Table 2.

Table 3
Diagnostic performance of magnetic resonance imaging in the present study

Variable Myoma Adenomyosis Sarcoma

Patients 119/135 26/135 5/135
Disease incidence, % 88.15 19.26 3.70
Sensitivity 94.12 46.15 60.00
Specificity 68.74 99.08 99.23
Positive likelihood ratio 3.01 50.31 78.00
Negative likelihood ratio 0.09 0.54 0.40
PPV 95.73 92.31 75.00
NPV 61.11 88.52 98.47
AUC (95% CI) 0.81 (0.74–0.88)a 0.73 (0.64–0.80)a 0.80 (0.72–0.86)a

AUC 5 area under the curve; CI 5 confidence interval; PPV 5 positive predictive value; NPV 5 negative predictive value.
a
p , .001.
624 Journal of Minimally Invasive Gynecology, Vol 19, No 5, September/October 2012

The Cramer V indicator for agreement between MRI and Fig. 1

histologic diagnosis was 0.67 (p , .001), and Cohen k coef-
Receiving operating characteristic curves of the diagnostic performance
ficient for agreement between MRI and histologic diagnosis
of magnetic resonance imaging in diagnosis of myomas (A) and adeno-
was 0.61 (SE, 0.07; p , .001); both indicate a statistically
myosis (B).
significant relationship.
The sensitivity, specificity, and PPV and NPV predictive
a of MRI in the diagnosis of myomas, adenomyosis, and
uterine sarcomas are given in Table 3.
The diagnostic performance of MRI using the histologic
results as the criterion standard was measured using ROC
analysis. The AUC for the value of MRI in the diagnosis
of myomas and adenomyosis are given in Table 3, and the
ROC curves are shown in Figure 1.

Discussion
The results of the present study highlight the value of
MRI in the diagnosis of myomas and adenomyosis of the
uterus. The most important finding is that after the construc-
tion of ROC curves, the AUC is high, 0.81 for the diagnosis
of myomas and 0.73 for adenomyosis. Moreover, it seems
that MRI is a diagnostic method characterised primarily by
a high PPV in the diagnosis of myomas (95.7%) and adeno-
myosis (92.3%).
The significance of our results is apparent in everyday
clinical practice. Myomas are common conditions that usu-
ally do not demand treatment [30]. Nevertheless, under spe-
cific conditions (when myomas become symptomatic),
patients seek gynecologic intervention. The options for my-
oma treatment are numerous, ranging from total abdominal
hysterectomy to MR-guided focused US ablation. Appropri-
ate selection of patients for medical treatment, noninvasive
procedures (e.g., uterine artery embolization or MR-guided
focused US ablation), or surgery relies on accurate assess-
ment of the size, number, location, and perfusion of the
myomas [31]. The evaluation of the proximity of the myo-
mas to the uterine cavity might help the gynecologist to
plan the operation and avert missing myomas during surgery
[32,33]. A diagnostic test with increased PPV in the
diagnosis of myomas matches the purposes of clinicians,
uses resources in the best possible way, and increases cost-
effectiveness. Our results indicate that the possibility of
a myoma found at MRI to be correctly diagnosed is high
(95.7%). Considering that the sensitivity of MRI in the diag-
nosis of myomas is also high (94.1% in the present study), [12,17–19]. The ROC results of our study are in full
MRI seems to fulfill significant clinical criteria for the inves- agreement with these studies, confirming the high
tigation of this condition. Other studies in which MRI has diagnostic skill of MRI in the detection of myomas (Table 4).
been used for the diagnosis of myomas have reported similar Similarly, the results of our study indicate that MRI dem-
results (Table 4). onstrates a high diagnostic performance in adenomyosis.
The present study also used ROC analysis to measure the This frequently occurring condition causes substantial mor-
overall clinical performance of MRI in the diagnosis of bidity. Compared with myomas, adenomyosis is a more ob-
myomas [15]. This comparison has not been calculated in scure disease, both in terms of diagnosis and treatment. Until
the literature previously. Thus, we created ROC curves based recently, the criterion standard for a definitive diagnosis of
on data from previous relevant published studies in which adenomyosis was histopathologic analysis of hysterectomy
MRI was used in the diagnosis of myomas. The AUCs cre- specimens [34,35]. However, a recent meta-analysis has
ated from this calculation ranged from 0.60 to 0.99 demonstrated that both US and MRI may enable accurate
Stamatopoulos et al. MRI in Adenomyosis and Uterine Myomas 625

Table 4
Diagnostic performance of MRI in the present study vs the literature

No. of Incidence
Variable patients Study design of disease Sensitivity Specificity PPV NPV AUC (95% CI)
Myoma
Present studya 135 Prospective 88.15 94.12 68.74 95.73 61.11 0.81 (0.74–0.88)b
Levens et al [20], 2009c 18 Prospective 92.64 80.13 NA 90.98 NA 0.60 (0.52–0.68)b
Moghadam et al [19], 2006a 153 Retrospective 94.12 94.44 33.33 95.77 27.27 0.98 (0.93–0.99)b
Dueholm et al [18], 2002a 106 Prospective 68.87 98.63 90.91 96.00 96.77 0.95 (0.89–0.98)b
Weinreb et al [17], 1990a 19 Prospective 73.68 78.57 60.00 84.62 50.00 0.69 (0.44–0.88)
Adenomyosis
Present study 135 Prospective 19.26 46.15 99.08 92.31 88.52 0.73 (0.64–0.80)b
Moghadam et al [19], 2006 153 Retrospective 20.26 38.71 90.98 52.17 85.38 0.65 (0.57–0.72)b
Dueholm et al [24], 2001 106 Prospective 20.75 63.64 88.10 58.33 90.24 0.76 (0.67–0.84)b
Bazot et al [23], 2001 120 Prospective 33.33 77.50 92.50 83.78 89.16 0.85 (0.77–0.91)b
Reinhold et al [22], 1996 119 Prospective 23.53 85.71 85.71 64.86 95.12 0.86 (0.78–0.91)b
Ascher et al [21], 1994 20 Prospective 85.00 88.24 66.67 93.75 50.00 0.76 (0.54–0.93)b

AUC 5 area under the curve; CI 5 confidence interval; NA 5 data not available; NPV 5 negative predictive value; MRI 5 magnetic resonance imaging; PPV 5 positive
predictive value.
a
Diagnostic performance of MRI per person.
b
p , .001.
c
Diagnostic performance of MRI per lesion.

noninvasive diagnosis [36]. Medical or conservative surgical
treatment of adenomyosis has not been standardized, and the
efficacy of these options has been mostly ineffective in the
long term [37,38]. Therefore, the primary treatment
remains hysterectomy, an intervention associated with
multiple risks for the patient [10]. Thus, it is of paramount
importance to determine a noninvasive method for definitive
diagnosis of adenomyosis. Our results indicate that the pos-
sibility of adenomyosis found at MRI to be correctly diag-
nosed is high (92.3%). Considering that the specificity of
MRI in the diagnosis of adenomyosis is almost 100%
(99.1% in the present study), MRI seems to fulfill significant
clinical criteria for the investigation of adenomyosis. Other
studies in which MRI has been used for diagnosis of adeno-
myosis have reported similar results (Table 4).
In similar fashion similar as mentioned above, we created
ROC curves based on data from previous relevant published
studies in which MRI was used in diagnosis of adenomyosis.
The AUCs from these studies ranged from 0.65 to 0.85
[10,19,21–23]. ROC analysis of the MRI performance in
diagnosis of adenomyosis in our data showed that the AUC
was 0.73, which is in agreement with the published literature.
An interesting finding of the present study was the 5 cases
of uterine sarcoma. The incidence of uterine sarcoma may
seem high; however, this may be because the study was
performed in a tertiary referral center. Two cases of uterine
sarcoma were incorrectly diagnosed at MRI. Both were stro-
mal sarcomas; one case was misdiagnosed as myoma, and
another as endometrial carcinoma (Table 2). However, 4 of
5 sarcomas were diagnosed accurately as ‘‘cancer,’’ and
there was only 1 case in which MRI indicated sarcoma
that was histologically diagnosed as myoma. Thus, the sen-
sitivity of MRI was above average (60%), the specificity was
high (99%), and the NPV was high (98.5%) in diagnosis of
sarcomas. Because of the small number of uterine sarcomas
in the present study and because the diagnosis of sarcomas
using MRI was not the primary objective of the study, no ro-
bust conclusion can be reached about the clinical usefulness
of MRI in diagnosis of uterine sarcomas. Larger, prospective
studies are required to answer this clinical problem because
it is important to differentiate cancer from benign myomas
so that appropriate consultation and management can be per-
formed before treatment.
The strength of the present study is associated with multi-
ple factors. The calculation of ROC curves for the diagnostic
performance of MRI in the investigation of myomas and ad-
enomyosis has not been an objective of previous studies.
However, with the use of ROC analysis, clinical decision
making is orientated toward safer, more cost-effective diag-
nostic tools that minimize patient discomfort. ROC curves
based on the data from the present study suggest that MRI
is a useful diagnostic tool both for myomas and adenomyosis.
The use of histopathology as the criterion standard for di-
agnosis of myomas and adenomyosis is an important point.
The method of the study warranted the pathologic result to
confirm the final diagnosis so that diagnoses made using
conservative methods (US, MRI, computed tomography, or
combinations of these methods), which are inherently faulty,
do not hamper the results.
The prospective design of the study and the recruitment
of consecutive patients, along with the blinding of indepen-
dent observers to the findings, and adequate power of the
study to show statistical significance of any differences
found, empower the final results.
626
In conclusion, MRI exhibits a high AUC for diagnosis of
both adenomyosis and myomas. The PPVof MRI in diagnosis
of adenomyosis and uterine myomas is high as well. There-
fore, MRI seems to be a useful technique in everyday clinical
practice in the diagnostic approach of these common
conditions. Clinicians may use MRI to increase diagnostic
accuracy, select appropriate treatment, individualize man-
agement for each patient, and improve therapeutic rates.
References
1. Blake RE. Leiomyomata uteri: hormonal and molecular determinants
of growth. J Natl Med Assoc. 2007;99:1170–1184.
2. Okolo S. Incidence, aetiology and epidemiology of uterine fibroids.
Best Pract Res Clin Obstet Gynaecol. 2008;22:571–588.
3. Laughlin SK, Schroeder JC, Baird DD. New directions in the epidemi-
ology of uterine fibroids. Semin Reprod Med. 2010;28:204–217.
4. Taran FA, Brown HL, Stewart EA. Racial diversity in uterine
leiomyoma clinical studies. Fertil Steril. 2010;94:1500–1503.
5. Vitiello D, McCarthy S. Diagnostic imaging of myomas. Obstet
Gynecol Clin North Am. 2006;33:85–95.
6. Grimbizis GF, Assimakopoulos E, Chatzigeorgiou KN, Vavilis D,
Miliaras D, Bontis JN. Successful laparoscopic treatment of a primary
peritoneal leiomyoma. Acta Obstet Gynecol Scand. 2006;85:1399–1400.
7. Theodoridis TD, Zepiridis L, Grimbizis G, Bontis J. Laparoscopic man-
agement of broad ligament leiomyoma. J Minim Invasive Gynecol.
2005;12:469.
8. Theodoridis TD, Zepiridis L, Mikos T, et al. Comparison of diagnostic
accuracy of transvaginal ultrasound with laparoscopy in the manage-
ment of patients with adnexal masses. Arch Gynecol Obstet. 2009;
280:767–773.
9. Grimbizis GF, Mikos T, Zepiridis L, et al. Laparoscopic excision of
uterine adenomyomas. Fertil Steril. 2008;89:953–961.
10. Farquhar C, Brosens I. Medical and surgical management of adenomyo-
sis. Best Pract Res Clin Obstet Gynaecol. 2006;20:603–616.
11. Jeng CJ, Huang SH, Shen J, Chou CS, Tzeng CR. Laparoscopy-guided
myometrial biopsy in the definite diagnosis of diffuse adenomyosis.
Hum Reprod. 2007;22:2016–2019.
12. Levens ED, Potlog-Nahari C, Armstrong AY, et al. CDB-2914 for uter-
ine leiomyomata treatment: a randomized controlled trial. Obstet
Gynecol. 2008;111:1129–1136.
13. Rajan DK, Margau R, Kroll RR, et al. Clinical utility of ultrasound ver-
sus magnetic resonance imaging for deciding to proceed with uterine
artery embolization for presumed symptomatic fibroids. Clin Radiol.
2011;66:57–62.
14. Benagiano G, Brosens I, Carrara S. Adenomyosis: new knowledge is
generating new treatment strategies. Womens Health (Lond Engl).
2009;5:297–311.
15. Nirula R, Basel K. Determining and interpreting the accuracy of a test.
Curr Surg. 2006;63:190–193.
16. Miller AB. Screening. In: Ahrens W, Pigeot I, editors. Handbook of
Epidemiology. Berlin, Germany: Springer-Verlag; 2005. p. 1267–1304.
17. Weinreb JC, Barkoff ND, Megibow A, Demopoulos R. The value of
MR imaging in distinguishing leiomyomas from other solid pelvic
masses when sonography is indeterminate. AJR Am J Roentgenol.
1990;154:295–299.
18. Dueholm M, Lundorf E, Hansen ES, Ledertoug S, Olesen F. Accuracy
of magnetic resonance imaging and transvaginal ultrasonography in the
Journal of Minimally Invasive Gynecology, Vol 19, No 5, September/October 2012
diagnosis, mapping, and measurement of uterine myomas. Am J Obstet
Gynecol. 2002;186:409–415.
19. Moghadam R, Lathi RB, Shahmohamady B, et al. Predictive value of
magnetic resonance imaging in differentiating between leiomyoma
and adenomyosis. JSLS. 2006;10:216–219.
20. Levens ED, Wesley R, Premkumar A, Blocker W, Nieman LK. Mag-
netic resonance imaging and transvaginal ultrasound for determining
fibroid burden: implications for research and clinical care. Am J Obstet
Gynecol. 2009;200:537.
21. Ascher SM, Arnold LL, Patt RH, et al. Adenomyosis: prospective com-
parison of MR imaging and transvaginal sonography. Radiology. 1994;
190:803–806.
22. Reinhold C, McCarthy S, Bret PM, et al. Diffuse adenomyosis: compar-
ison of endovaginal US and MR imaging with histopathologic correla-
tion. Radiology. 1996;199:151–158.
23. Bazot M, Cortez A, Darai E, et al. Ultrasonography compared with
magnetic resonance imaging for the diagnosis of adenomyosis: correla-
tion with histopathology. Hum Reprod. 2001;16:2427–2433.
24. Dueholm M, Lundorf E, Hansen ES, Sørensen JS, Ledertoug S,
Olesen F. Magnetic resonance imaging and transvaginal ultrasonogra-
phy for the diagnosis of adenomyosis. Fertil Steril. 2001;76:588–594.
25. Devlieger R, D’Hooghe T, Timmerman D. Uterine adenomyosis in the
infertility clinic. Hum Reprod Update. 2003;9:139–147.
26. Kido A, Togashi K, Koyama T, Yamaoka T, Fujiwara T, Fujii S. Dif-
fusely enlarged uterus: evaluation with MR imaging. Radiographics.
2003;23:1423–1439.
27. Rha SE, Byun JY, Jung SE, et al. CT and MRI of uterine sarcomas and
their mimickers. AJR Am J Roentgenol. 2003;181:1369–1374.
28. Peric H, Fraser IS. The symptomatology of adenomyosis. Best Pract
Res Clin Obstet Gynecol. 2006;20:547–555.
29. Gilks CB, Clement PB, Hart WR, Young RH. Uterine adenomyomas
excluding atypical polypoid adenomyomas and adenomyomas of endo-
cervical type: a clinicopathologic study of 30 cases of an underempha-
sized lesion that may cause diagnostic problems with brief
consideration of adenomyomas of other female genital tract sites. Int
J Gynecol Pathol. 2000;19:195–205.
30. Divakar H. Asymptomatic uterine fibroids. Best Pract Res Clin Obstet
Gynaecol. 2008;22:643–654.
31. Parker WH. Etiology, symptomatology, and diagnosis of uterine
myomas. Fertil Steril. 2007;87:725–736.
32. Grimbizis GF, Tsolakidis D, Mikos T, et al. A prospective comparison
of transvaginal ultrasound, saline infusion sonohysterography, and
diagnostic hysteroscopy in the evaluation of endometrial pathology.
Fertil Steril. 2010;94:2720–2725.
33. Dueholm M, Lundorf E, Olesen F. Imaging techniques for evaluation of
the uterine cavity and endometrium in premenopausal patients before
minimally invasive surgery. Obstet Gynecol Surv. 2002;57:388–403.
34. Peric H, Fraser IS. The symptomatology of adenomyosis. Best Pract
Res Clin Obstet Gynaecol. 2006;20:547–555.
35. Vercellini P, Vigano P, Somigliana E, Daguati R, Abbiati A, Fedele L.
Adenomyosis: epidemiological factors. Best Pract Res Clin Obstet
Gynaecol. 2006;20:465–477.
36. Champaneria R, Abedin P, Daniels J, Balogun M, Khan KS. Ultrasound
scan and magnetic resonance imaging for the diagnosis of adenomyo-
sis: systematic review comparing test accuracy. Acta Obstet Gynecol
Scand. 2010;89:1374–1384.
37. Rabinovici J, Stewart EA. New interventional techniques for adeno-
myosis. Best Pract Res Clin Obstet Gynaecol. 2006;20:617–636.
38. Levgur M. Therapeutic options for adenomyosis: a review. Arch
Gynecol Obstet. 2007;276:1–15.