Radiology Section DOI : 10.7860/JCDR/2017/26030.10182
Year : 2017 | Month : Jul | Volume : 11 | Issue : 7 Page : TC01 - TC05

Pelvic Mass Lesions in Females: Tissue Characterization Capability of MRI

Pramod Ramchand Shaha1, Rahul Khetawat2, Kulamani Sahoo3, Ashish Garg4, Mohd. Abbas Ilyas5, Gaurav Khairnar6, Sahil Garg7, Shree Hari Budgemwar8

1 Professor, Department of Radiology, Krishna Institute of Medical Sciences, Karad, Maharashtra, India.
2 Resident, Department of Radiology, Krishna Institute of Medical Sciences, Karad, Maharashtra, India.
3 Professor and HOD, Department of Radiology, Krishna Institute of Medical Sciences, Karad, Maharashtra, India.
4 Senior Resident, Department of Radiology, MAMC and GB Pant Hospital, New Delhi, India.
5 Resident, Department of Radiology, Krishna Institute of Medical Sciences, Karad, Maharashtra, India.
6 Resident, Department of Radiology, Krishna Institute of Medical Sciences, Karad, Maharashtra, India.
7 Resident, Department of Radiology, MMU, Maulana, Haryana, India.
8 Resident, Department of Radiology, Krishna Institute of Medical Sciences, Karad, Maharashtra, India.


NAME, ADDRESS, E-MAIL ID OF THE CORRESPONDING AUTHOR: Dr. Rahul Khetawat, Resident, Department of Radiology, Krishna Institute of Medical Sciences, Karad-415539, Maharashtra, India.
E-mail: rahul.med11@gmail.com
Abstract

Introduction

Magnetic Resonance Imaging (MRI) of adult female pelvis is a well-established tool in the evaluation of utero-ovarian lesions and is often used to supplement ultrasonography. The need for diagnostic surgical intervention has largely been eclipsed with the advent of MRI, which has become the imaging modality of choice for characterization of pelvic masses.

Aim

To assess the role of MRI in female pelvic mass lesions and to exploit the tissue characterization capability of MRI.

Materials and Methods

A prospective observational study was done on all patients referred to Department of Radiodiagnosis, Krishna Hospital, Karad, for MRI pelvis with clinically suspected uterine and adnexal masses or with indeterminate diagnosis on ultrasonography. Study was conducted between September 2014 to August 2016 with a sample size of 100 patients. All patients were scanned using 1.5 Tesla Seimens Avanto (Tim+Dot) scanner with Body matrix coil Tim. Histopathology was taken as gold standard. Results on continuous measurements were presented on Mean±SD (Min-Max) and results on categorical measurements were presented in Number (%).

Results

Among 100 cases, on MRI, the maximum number of patients was having uterine lesions (48) followed by ovarian lesions (40), inconclusive adnexal/ovarian lesions (6), adnexal lesions (4). Two patients had normal findings. This correlated well with histopathology results, which showed the maximum number of patients were having uterine lesions (48) followed by ovarian lesions (41), adnexal lesions (5). Normal findings were observed in two patients.

Conclusion

Due to excellent depiction of pelvic anatomy, non-invasiveness and absence of ionizing radiation, MRI is an excellent tool for assessment of utero-ovarian disorders, for detecting and characterization of various diseases, and staging patients with carcinomas where accurate diagnosis will make an impact on their surgical and medical management planning.

Introduction

Magnetic Resonance Imaging (MRI) is an imaging modality that has been developed and used since mid-1970s [1]. MRI has several advantages over computed tomography and ultrasonography.

One important feature is its non-invasiveness. A second feature that makes MRI particularly attractive is its capability for multiplanar imaging. A third advantage of MRI is its excellent tissue-differentiating capabilities. A fourth advantage of MRI is its intrinsic sensitivity to flowing blood. Both arterial and venous abnormalities can be assessed by MRI [1].

The indications for MRI for utero-ovarian lesions have expanded considerably over the past decade. The impetus behind these expanding indications is multifactorial. First, there has been widespread dissemination of MR hardware and software techniques that allow the routine acquisition of images with high spatial or temporal resolution. Second, the advent of minimally invasive therapies in the management of gynaecologic disorders has created a need for increased accuracy in the preoperative evaluation of these patients, Finally, in this era of cost containment, several studies have shown that the appropriate use of MR imaging in the diagnostic algorithm minimizes cost [2]. Some of the indications for MRI evaluation of the female genital tract are: infertility, congenital anomalies [3], ovarian and cervical lesions, pelvic inflammatory disease etc.

MRI of the adult female pelvis is a well-established tool in the evaluation of utero-ovarian lesions and is often used as a supplement to ultrasonography. The need for diagnostic surgical intervention has largely been eclipsed with the advent of MRI, which has become the imaging modality of choice for characterization of congenital Mullerian anomalies [4].

MRI is gaining momentum for staging gynaecologic malignancies. Ovarian cancer is the second most common pelvic tumour and the leading cause of death from gynaecologic malignancy. In patients with ovarian cancer, MRI determination of disease extent helps treatment planning either as a surgical roadmap or to identify non resectable patients. MRI staging is adjunct to clinical and surgical staging in women with cervical or endometrial cancer, respectively [5,6].

Therefore, an adnexal mass that appears suspicious at USG may be correctly diagnosed as a benign lesion at MR, preventing inappropriate radical surgery. MR is the most useful modality for evaluating adnexal lesions that are indeterminate at gray-scale ultrasound. Therefore, the study was carried out to assess the role of MRI in female pelvic mass lesion and to exploit the tissue characterization capability of MRI with few secondary aims like to study the spectrum of diverse nature of pelvic mass lesions, to determine the origin, tissue content and characterization of sonographically inconclusive uterine and adnexal masses, and to differentiate utero-ovarian lesions as neoplastic and non-neoplastic.

Materials and Methods

A prospective observational study was done in Department of Radiodiagnosis, Krishna Hospital, Karad, Maharashtra, India, for MRI pelvis with clinically suspected uterine and adnexal masses or with indeterminate diagnosis on ultrasonography. Over the study period of two years, from September 2014 to August 2016, 128 patients were referred for MRI pelvis, out of which 100 patients fulfilled our inclusion criteria and formed the sample size. Study was conducted after taking the institutional ethical clearance. Inclusion criteria were all patients with clinically suspected uterine and adnexal masses or with indeterminate diagnosis on sonography, non-pregnant females and patients of all age groups. Exclusion criteria were pregnant females, patients having any metallic implants, patients having history of claustrophobia, and all patients who do not consent to be a part of the study.

Ultrasonography pelvis and MRI pelvis of all the patients was done after obtaining the written informed consent in English as well as in regional language (Marathi) and imaging was done as per imaging protocol and images were stored in compact disc. All patients in our study underwent pelvic ultrasonography (SIEMENS ACUSON x300) by using both curvilinear (3-5 MHz) and linear (6-9 MHz) probes, with the patient in supine position. This is followed by pelvic MRI examination on a 1.5 Tesla scanner (SIEMENS TIM + DOT AVANTO). Scanning was performed after a fasting period of four to six hour prior to the examination to limit bowel motion. Imaging was in supine position with a Body matrix coil # Tim placed on the torso covering the entire imaging area. Initially a localizer scan was performed in transverse, sagittal and coronal orientation. This was followed by T1W/TSE {TE=10 ms / TR=500 ms} in sagittal planes; T2W/TSE {TE=80-90 ms / TR=4000-4200 ms} - axial, coronal and sagittal planes; STIR {TE=30-40 ms / TR=3500-4000 ms} -coronal planes, T1FS {TE=10 ms / TR=500 ms} – coronal and axial planes and T2FS {TE=80-90 ms / TR=4000-4200 ms} in axial and sagittal planes as required.

Gadolinium dimeglumine (Omniscan) 10 ml was administered as and when required. Nearly, 10 ml saline flush was also given after administration of gadodiamide. Post contrast T1FS {TE=10 ms/ TR=500 ms} images were obtained in axial, coronal and sagittal planes. Both gadolinium dimeglumine and normal saline were administered at the rate of 2 ml/sec. The parameters included matrix size of 192–256 × 256–512, field of view (appropriate to body habitus), 230–250 mm for the body coil, and 3-4 mm section thickness with an intersection gap of 30%-40%. Post processing was done by inbuilt software.

Both ultrasonography and MRI were performed by same group of researchers. In MRI size, shape, signal intensity, character, and postcontrast enhancement of the lesions were observed. After the MRI it was the decision of the referring physician to operate or not, and whenever the patients were operated biopsy was done. In our study, four patients were not operated.

Statistical Analysis

The statistical software namely MedCalc 9.0.1, Systat 12.0 were used for the analysis of the data and Microsoft word and Excel have been used to generate graphs, tables etc. Descriptive and inferential statistical analysis has been carried out in the present study. Results on continuous measurements are presented on Mean±SD (Min-Max) and results on categorical measurements are presented in Number (%).

Results

The maximum number of patients were in the age group 51-60 (23) followed by 31-40 (18) and 20-30 (16). The range was between 10 and 80 years.

The maximum cases in USG diagnosis group were those of bulky uterus (10) followed by fibroid uterus (8) and bulky cervix + uterus (8). Distribution of lesions on USG, MRI and histopathology are mentioned in [Table/Fig-1].

Distribution of lesions – on USG, MRI, histopathology.

LesionUSGMRIHistopathology
Uterine484848
Ovarian254041
Adnexal-45
Inconclusive276-
Normal Study/NA-26
Total100100100

The maximum cases in MRI diagnosis group were those of fibroid uterus (22=46% of the uterine lesions) followed by endometrial carcinoma (10=21% of the uterine lesions) and carcinoma cervix (10=21% of the uterine lesions) [Table/Fig-1].

The maximum number of patients with ovarian lesions were having serous cystadenoma (right) n-6 (19.36% of ovarian lesions) and serous cystadenoma (left) n-6 (19.36% of ovarian lesions) followed by mucinous cystadenoma (right) n-5 (16.13% of ovarian lesions), mucinous cystadenoma (bilateral) n-3 (9.68% of ovarian lesions).

Internal and external Characteristics of the lesions on MRI were described in the [Table/Fig-2,3,4,5,6 and 7].

Well-defined/ill-defined - distribution of lesions studied.

LesionWell-DefinedIll-DefinedNot Applicable
Uterine24242
Ovarian/Adnexal4640
Total70282

Solid/cystic - distribution of lesions studied.

Fat planes - distribution of lesions studied.

LesionsPreservedNot PreservedNot Applicable
Uterine4442
Ovarian/Adnexal4820
Total9262

Mass effect - distribution of lesions studied.

Mass EffectPresentAbsentNot Applicable
Uterine5432
Ovarian/Adnexal20300
Total25732

Gadolinium enhancement - distribution of lesions studied.

Internal characteristics of the ovarian/adnexal lesions.

CharacteristicsEpithelial cell tumoursOvarian/adnexal lesions
SCAMCAMCACSCACRO/ACCMO/AMLT-OCLLOCCDER MOID
Nodules42
Fat2
Hemorrhage2222
Calcifications22
Septations10821412

ISCA - Serous Cystadenoma; MCA - Mucinous Cystadenoma; SCAC - Serous Cystadenocarcinoma; MCAC - Mucinous Cystadenocarcinoma; RO/ACC - Right ovarian/adnexal cystic lesion; MO/A ML - Malignant ovarian/adnexal mass lesion; T-OCL - Tubo-ovarian cystic lesion; LOCC - Left ovarian complex cyst.


Neoplastic i.e., benign vs malignant and non-neoplastic distribution of lesions are mentioned in [Table/Fig-8,9].

Neoplastic/non-neoplastic - distribution of lesions studied.

LesionsNeoplasticNon NeoplasticNot Applicable
Uterine4802 (Normal study)
Ovarian/Adnexal4820
Total9622

Benign/Malignant - distribution of lesions studied by MRI study.

LesionsBenignMalignantNot Applicable
Uterine28202 (Normal study)
Ovarian/Adnexal4280
Total70282

Discussion

MRI is an excellent tool in the tissue characterization of female pelvic mass lesions, differentiating non-neoplastic lesions from neoplastic lesions and benign from malignant lesions, apart from determining origin of the lesion in case of indeterminate diagnosis on ultrasonography.

A] Uterine Lesions

A1) Uterine fibroids [Table/Fig-10]: In the present study, among the uterine lesions uterine fibroids, only 16 (n=16→33%) were diagnosed with USG while 22 (n=22→46%) were diagnosed by MRI. This was similar to studies by Hamlin JD et al., where MR imaging to be more accurate than USG [7]. Two of the lesions were given a doubtful diagnosis of bulky cervix which was proved to be a uterine leiomyoma on HPE. In another two cases, which were diagnosed on USG as carcinoma endometrium turned out to be a uterine fibroid on MRI.

a) T2WI sagittal; b) T2FS sagittal; c) T1FS sagittal contrast images - showing altered signal intensities in posterior myometrium and displacing the endometrium anteriorly - Submucosal fibroid (arrows).

A2) Endometrial carcinoma [Table/Fig-11]: MRI could diagnose only 10 (n=10→21%) endometrial carcinomas of all uterine lesions. two of which turned to be uterine leiomyoma on HPE.

a) T2W sagittal; b) T2W Axial c) T1FS Sagittal contrast images - Well defined diffuse circumferential lesion causing thickening of the endometrium involving anterior and posterior endometrial wall showing heterogenous enhancement (less enhancement than normal myometrium) with interruption of junctional zone indicating myometrial invasion (>50%) – CA endometrium (arrows).

Patients typically present between the fifth and eighth decades of life, with a mean age of presentation was 64 years. The most common clinical manifestation was abnormal vaginal bleeding. The differential diagnosis for endometrial cancer includes cervical cancer, endometrial polyps, and mesenchymal tumours of the uterus. Distinguishing between these entities is easier in smaller tumours [8,9].

A3) Adenomyosis [Table/Fig-12]: There were four cases of adenomyosis (n=4→8%) of all uterine lesions. Two of the lesions appeared as ill-defined ovoid/diffuse region of thickening with small high T2 signal regions whereas the other lesions had a striated appearance. However, on HPE the lesion with striated appearance turned out to be normal proliferative endometrium.

a) T2W sagittal; b) T2W axial; c) Bulky uterus with thick endometrium with endometrial collection. Both anterior and posterior myometrium shows multiple tiny haphazardly distributed T2 hyperintense areas. Endo-myometrial borders are hazy – Adenomyosis (arrow).

MR imaging is also an accurate, noninvasive modality for diagnosing adenomyosis with a high sensitivity (78%–88%) and specificity (67%–93%) – according to a study done by Reinhold C et al., [10].

A4) Carcinoma cervix: MR imaging could diagnose 10 (n=10→21%) cases of carcinoma cervix of all uterine lesions.

Patients typically present between the fouth and seventh decades of life, with a mean age at presentation of 56 years. The mean age was 53 years according to a study by Yamashita Y et al., [11]. Another study by Reinhold C et al., and McCarthy S et al., showed the mean age to be 45 years [12,13].

According to a study by Szklaruk J et al., where they had a maximum of squamous cell carcinomas – 90%, followed by adeno-squamous carcinoma [14]. Appearance and extension of each lesion vary according to the stage of the lesion, ranging from stage IB (six cases) to IIIA (one case).

B] Ovarian and Inconclusive Ovarian/Adnexal Lesions

Out of the 50 ovarian and adnexal lesions, origin of 27 (54%) lesions could not be made out by USG, whereas origin of only 6 (12%) lesions could not be made out and 44 (88%) lesions could be made out by MRI. Inconclusive lesions were later proved to be three ovarian and three adnexal in origin by HPE. According to a similar study done by Dwivedi AN et al., they showed that MRI was 97.7% sensitive and 73.1% specific for diagnosing mass of ovarian origin [15].

Out of the 50 ovarian and adnexal lesions, 23 lesions were right sided, 15 left sided, eight were bilateral. The laterality of 4 lesions could not be made out because of the large size of the tumours which were crossing the midline and as the ovaries could not be seen separately.

Out of the 50 ovarian and adnexal lesion 48 (96%) were neoplastic and 2 (4%) were non-neoplastic on MRI. Our findings were more or less similar to a study done by Sohaib SA et al., which shows 91%-93% overall accuracy for differentiating benign from malignant tumours [16].

B1) Epithelial tumours: There were 28 (58%) epithelial tumours of the 50 ovarian/adnexal lesions. There were 12 serous cystadenomas (six patients had right and six patients had left serous cystadenoma) [Table/Fig-13] which comprise 25% of the ovarian/adnexal neoplasm studied.

a) T2W sagittal; b) T2W axial; c) T1FS axial - A large well defined solid cystic lesion in pelvis with enhancing multiple septations (arrow) within it – Serous cystadenoma.

Serous cystadenomas comprise approximately 20% of the benign ovarian masses. They present as complex thin-walled, unilocular- or multilocular cysts, with variable sizes, sometimes achieving more than 20 cm. The image of its interior shows echogenic contents, possibly also revealing areas of papillary projections [17].

In our study there were 10 (21%) mucinous cystadenomas out of the 48 neoplastic ovarian lesions. This was similar to a study done by Seidman JD et al., mucinous ovarian tumours are less common than serous neoplasms. They represent 20% of all ovarian tumours and approximately 10% of all malignant ovarian tumours [18].

Two cases were mucinous cystadenocarcinoma [Table/Fig-14] which comprises 4% of the ovarian neoplasms in the present study.

a) T2W sagittal; b) T2FS sagittal; c) T1FS sagittal contrast images- Left adnexa shows large lobulated cystic lesion showing variable signal intensity and peripheral enhancement with papillary projections (arrow) and thick enhancing septations (~5 mm) - mucinous cyst adenocarcinoma.

B2) Germ cell tumours: We found two cases of dysgerminoma during the course of the study. The patients were 11 and 13-year-old. Most cases occur in adolescence and early adulthood, although it may occur at any age according to Shaaban AM et al., [19].

MR imaging revealed a solid mass divided into lobules by fibrovascular septa. The septae showed intense enhancement after administration of contrast material. This finding corroborates with Shaaban AM et al., who state that dysgerminomas have low signal intensity relative to muscle on T1- weighted images and are isointense or slightly hyperintense on T2-weighted images [19]. Usually, septa are hypointense isointense on T2-weighted images and are difficult to appreciate on T1-weighted images, with intense enhancement after administration of contrast material.

There were four patients who had dermoid. Two patients had bilateral dermoid. Caruso PA et al., found that 10% of dermoid were bilateral [20]. Three of the four dermoid were unilocular (75%) and it corroborates with Outwater EK et al., who states that 88% of benign cystic teratomas are unilocular [21].

At MR imaging, the sebaceous component of the dermoid cysts had very high signal intensity on T1-weighted images similar to that of retroperitoneal fat. The signal intensity of the sebaceous component on T2-weighted images was variable, usually near that of fat. There was complete drop in signal intensity on fat sat sequences. This correlates with study done by Guinet C and Kido A et al., [22,23].

Limitation

Potential problems in MRI detection of ovarian lesions include difficulty in determining the adnexa of origin of a large mass. The organ of origin is also difficult to determine in larger adnexal lesions. Secondly, the duration of the study was small so few cases of each tumour type could be collected. A longer duration study with larger number of cases would show a better picture. Thirdly, chances are that only ultrasonographically complex lesions were referred for MR examination which could also have introduced some sort of bias. It is also possible that cases with disseminated uterine and ovarian malignancy were not referred for MR evaluation.

Conclusion

In conclusion it can be said that due to excellent depiction of pelvic anatomy and absence of ionizing radiation, MRI is an excellent tool for the assessment of disorders of the uterus and ovary. MRI is a better modality for detecting, characterization of various diseases, staging patients with carcinomas where accurate diagnosis will make an impact on their surgical and medical management planning in comparison with USG.

ISCA - Serous Cystadenoma; MCA - Mucinous Cystadenoma; SCAC - Serous Cystadenocarcinoma; MCAC - Mucinous Cystadenocarcinoma; RO/ACC - Right ovarian/adnexal cystic lesion; MO/A ML - Malignant ovarian/adnexal mass lesion; T-OCL - Tubo-ovarian cystic lesion; LOCC - Left ovarian complex cyst.

References

[1]Kay HH, Spritzer CE, Magnetic Resonance Imaging in Gynecology [Internet] 2004 Last accessed on Apr 27 20171(93)Lippincott Williams & WilkinsAvailable from: https://www.glowm.com/reis/glowm/cd/pages/v1/v1c093.html  [Google Scholar]

[2]Caroline R, Magnetic resonance imaging in gynecologic disease Topics in Magnetic Resonance Imaging 2003 14(4):267-68.  [Google Scholar]

[3]McLeary M, Kjellin IB, Kirk SR, Magnetic resonance imaging of the pediatric female pelvis: a pictorial essay Journal of Women’s Imaging 2001 3(1):38-44.  [Google Scholar]

[4]Troiano B, Robert N, Magnetic resonance imaging of mullerian duct anomalies of the uterus Topics in Magnetic Resonance Imaging in Gynecologic Disease 2003 14(4):269-79.  [Google Scholar]

[5]Pretorius ES, Outwater EK, Hunt JL, Siegelman ES, Magnetic resonance imaging of the ovary Topics in Magnetic Resonance Imaging 2001 12(2):131-46.  [Google Scholar]

[6]Ascher S, Susan M, Takahama Jha J, Reena C, Staging of gynecologic malignancies Topics in Magnetic Resonance Imaging 2001 12(2):105-29.  [Google Scholar]

[7]Halmin JD, Pettersson H, Fitzsimmons J, Morgan LS, MR imaging of uterine leiomyomas and their complications J Comput Assist Tomogr 1985 9:902-07.  [Google Scholar]

[8]Bristow RE, Endometrial cancer Curr Opin Oncol 1999 11:388-93.  [Google Scholar]

[9]Rose PG, Endometrial carcinoma N Engl J Med 1996 335:640-49.  [Google Scholar]

[10]Reinhold C, McCarthy S, Bret P, Mehio A, Atri M, Zakarian R, Diffuse adenomyosis: comparison of endovaginal US and MR imaging with histopathologic correlation Radiology 1996 199(1):151-58.  [Google Scholar]

[11]Yamashita Y, Baba T, Baba Y, Nishimura R, Ikeda S, Takahashi M, Dynamic contrast-enhanced MR imaging of uterine cervical cancer: pharmacokinetic analysis with histopathologic correlation and its importance in predicting the outcome of radiation therapy Radiology 2000 216(3):803-09.  [Google Scholar]

[12]Reinhold C, Gallix BP, Ascher SM, Semelka Rc, Ascher SM, Reinhold C, Uterus and cervix MRI of the abdomen and pelvis: a text atlas, New York 1997:585-660.  [Google Scholar]

[13]McCarthy S, Haricak H, Higgins CB, Hricak H, Helms CA, The uterus and vagina Magnetic resonance imaging of the body 1997 3rd edNew York, NYLippincott_Raven:761-814.  [Google Scholar]

[14]Szklaruk J, Tamm EP, Choi H, Varavithya V, MR imaging of common and uncommon large pelvic masses Radiographics 2003 23:403-24.  [Google Scholar]

[15]Dwivedi A, Jain S, Shukla RC, Jain M, Srivastava A, Verma A, MRI is a state of art imaging modality in characterization of indeterminate adnexal masses J Biomedical Science and Engineering 2013 (6):309-13.  [Google Scholar]

[16]Sohaib SA, Sahdev A, Troppen PV, Jacobs IJ, Reznek RH, Characterization of adnexal mass lesions on MRI American Journalism Review 2003 180:1297-304.  [Google Scholar]

[17]Andrade Neto F, Palma-Dias R, Costa FS, Ultrasonography of adnexal masses: imaging findings Radiol Bras 2011 44(1):59-67.  [Google Scholar]

[18]Siedman JD, Russell P, Kurman RJ, Kurman RJ, Surface epithelial tumours of ovary Blaustein’s pathology of female genital tract 2002 5th edNew YorkSpringer Verlag:791-904.  [Google Scholar]

[19]Shaaban AM, Rezvani M, Elsayes KM, Baskin H, Mourad A, Foster BR, Ovarian malignant germ cell tumours: cellular classification and clinical and imaging features Radiographics 2014 34(3):777-801.  [Google Scholar]

[20]Caruso PA, Marsh MR, Minkowitz S, Karten G, An intense clinicopathologic study of 305 teratomas of the ovary Cancer 1971 27:343-48.  [Google Scholar]

[21]Outwater E, Siegelman E, Hunt J, Ovarian teratomas: tumour types and imaging characteristics RadioGraphics 2001 21(2):475-90.  [Google Scholar]

[22]Guinet C, Ghossain MA, Buy JN, Malbec L, Hugol D, Truc JB, Mature cystic teratomas of the ovary: CT and MR findings Eur J Radiol 1995 20:137-43.  [Google Scholar]

[23]Kido A, Togashi K, Konishi I, Dermoid cysts of the ovary with malignant transformation: MR appearance AJR Am J Roentgenol 1999 172:445-49.  [Google Scholar]