| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Genitourinary Imaging |
1 From the Dept of Radiology, Jefferson Medical College and Thomas Jefferson University Hospital, Gibbon Bldg 3350AB, 111 S 11th St, Philadelphia, PA 19107 (A.B.K., R.J.W., D.G.M.); the Dept of Health Care Policy, Harvard Medical School, Boston, Mass (J.V.T., D.J.C., B.J.M.); the Dept of Radiology, Brigham and Women's Hospital and Harvard Medical School, Boston (C.M.C.T., S.G.S., D.L.B.); the Depts of Radiology (U.M.H., S.S., J.E.K.) and Pathology (R.J.K.), Johns Hopkins Hospital, Baltimore, Md; the Dept of Radiology, Hospital of the University of Pennsylvania, Philadelphia (P.H.A., E.S.S., B.G.C.); and the Dept of Radiology, University of Michigan Medical Center, Ann Arbor (R.L.B., I.R.F., J.H.E.). Received Jun 26, 1998; revision requested Aug 6; revision received Oct 15; accepted Dec 16. Supported in part by Public Health Service grant U01 CA59401 from the National Cancer Institute. Address reprint requests to A.B.K. (e-mail: alfred.b.kurtz@mail.tju.edu).
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Two hundred eighty women suspected to have ovarian cancer were enrolled in a prospective study before surgery. Doppler ultrasonography (US), computed tomography (CT), and magnetic resonance (MR) imaging were used to evaluate the mass; conventional US, CT, and MR imaging were used to stage spread.
RESULTS: All three modalities had high accuracy (0.91) for the overall diagnosis of malignancy. In the ovaries, the accuracy of MR imaging (0.91) was higher than that of CT and significantly higher than that of Doppler US (0.78). In the extraovarian pelvis and in the abdomen, conventional US, CT, and MR imaging had similar accuracies (0.87–0.95). In differentiation of disease confined to the pelvis from abdominal spread, the specificity of conventional US (96%) was higher than that of CT and significantly higher than that of MR imaging (88%), whereas the sensitivities of MR imaging (98%) and CT (92%) were significantly higher than that of conventional US (75%).
CONCLUSION: MR imaging is superior to Doppler US and CT in diagnosis of malignant ovarian masses. There is little variation among conventional US, CT, and MR imaging as regards staging.
Index terms: Computed tomography (CT), comparative studies, 852.1211, 852.12112, 852.12115 • Magnetic resonance (MR), comparative studies, 852.1214, 852.121415, 852.12143 • Ovary, neoplasms, 852.32 • Receiver operating characteristic curve (ROC) • Ultrasound (US), comparative studies, 852.1298 • Ultrasound (US), Doppler studies, 852.12984
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Bimanual pelvic examination and serum CA-125 levels have failed to allow consistent detection of ovarian malignancy. Because the sensitivities of these techniques are often below 50% (2–4), imaging modalities, particularly ultrasonography (US), computed tomography (CT), and magnetic resonance (MR) imaging, have become indispensable. US performed with transabdominal and endovaginal techniques has demonstrated accuracies of up to 80% in evaluation of ovarian masses; US is better in detection of masses than in diagnosis of malignancy (5–9). Abdominal spread has not been evaluated with US, to our knowledge. Spectral analysis of Doppler waveforms (Doppler US), which is often directed by color Doppler US, allows detection of tumor flow. Doppler US has not demonstrated consistency in diagnosis of malignancy (10–15). Studies of contrast material–enhanced CT and MR imaging have shown accuracies of almost 80% in diagnosis of cancer and 80%–90% in detection of abdominal spread (16–22).
In general, comparison studies of US and MR imaging have shown MR imaging to be better in differentiation of benign from malignant masses on the basis of accurate identification of fat and hemorrhage in benign tumors (23–27). In one study, US and CT performed equally well in detection of masses; CT was superior in diagnosis of malignancy (28).
To our knowledge, only one study has compared all three of these modalities in diagnosis of ovarian cancer and its spread (29). This study evaluated conventional US, nonenhanced CT, and low-field-strength MR imaging. In 75 patients with 29 cancers, these modalities demonstrated accuracies of only 60% or lower. Because of advances in technology, a larger study that evaluated state-of-the-art techniques was thought to be needed.
The Radiology Diagnostic Oncology Group performed a 3-year, five-institution, prospective study of nearly 300 patients with ovarian masses. We evaluated the results of Doppler and conventional US, CT, and MR imaging and compared their performances in diagnosis of a malignant ovarian mass and staging of its spread.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Women over 18 years of age were admitted to the study if they were suspected to have ovarian cancer on the basis of abnormal results of a pelvic physical examination and/or a preliminary pelvic US study. Abnormal results of a physical examination were defined as detection of an enlarged ovary or an adnexal mass, sometimes with satellite pelvic nodularity. Abnormal US results were defined as detection of a complex (noncystic) and/or solid ovarian or adnexal mass, which was 5 cm or larger along its longest axis in a premenopausal woman and of any size in a postmenopausal woman. Complex and solid characteristics included heterogeneous and/or hyperechoic internal echoes, a wall thickness of 3 mm or greater, wall nodularity, solid components, and internal septa at least 2 mm thick.
When the results of physical examination and pelvic US were equivocal or negative, an additional inclusion criterion could be detection of complex ascites (internal echoes, septa) at US. The serum CA-125 level was not an inclusion criterion but was recorded when available.
Each patient had to be enrolled in at least two and ideally all three of the imaging modalities within 4 weeks before pelvic-abdominal surgery. The protocol and consent forms were approved by the institutional review board at each hospital. For evaluation of the ovarian or adnexal abnormality, the performances of Doppler US, CT, and MR imaging in diagnosis of malignancy were compared. Conventional US could not be included in this analysis because abnormal results at conventional US were an admission criterion. For evaluation of spread into the extraovarian pelvis and the abdomen, the performances of conventional US, CT, and MR imaging were compared.
Women were excluded from the study if they could not provide informed medical consent or if they were not a candidate for or did not need complete pelvic-abdominal surgical exploration. Other exclusion criteria were pregnancy and prior pelvic-abdominal laparoscopy or surgery within 6 months of entry into the study.
Of the 310 patients considered for the study, 30 were excluded because they had medical contraindications (n = 19), did not need extensive surgery (n = 6), refused surgery (n = 4), or had no disease (n = 1). The remaining 280 patients made up the study group. They ranged in age from 19 to 82 years with a mean age of 52.0 years and a median age of 51.5 years.
Conventional and Doppler US Protocol
State-of-the-art commercially available US equipment was used. The systems used were the UltraMark 9 HDI and HDI 3000 platforms (Advanced Technology Laboratories, Bothell, Wash) and the model 128XP and 128 XP/10 platforms (Acuson, Mountain View, Calif). All machines had transabdominal and endovaginal probes with maximal frequencies of 2–5 MHz and 5–7 MHz, respectively. All machines had both color and pulsed Doppler capability. Whenever possible, Doppler US of the ovarian or adnexal abnormality was performed endovaginally; transabdominal Doppler US of the abnormality was reserved for studies with nondiagnostic results. Premenopausal women were scheduled for US within 8 days of the start of their menstrual cycles (30). The extraovarian pelvis was evaluated with combined endovaginal and transabdominal US, whereas the abdomen was analyzed transabdominally.
CT Protocol
The CT scanners used were the Somatom Plus and Plus-S units (Siemens Medical Systems, Iselin, NJ) and the 9800 Advantage and Hi Speed Advantage units (GE Medical Systems, Milwaukee, Wis). The scanners were used in a dynamic or spiral mode with 2-second scanning times and suspended respiration. After opacification of the gastrointestinal tract, the pelvis was examined during peak arterial enhancement after the start of an injection of 150 mL of 60% contrast material (ionic or nonionic). The pelvis was scanned with 5-mm collimation; spiral CT was performed with a 5 mm/sec table speed and 5-mm reconstruction thickness, and incremental CT was performed with contiguous scans obtained at 10 scans per minute. In the abdomen, spiral CT and incremental CT were performed with 5-mm collimation at 8–10-mm intervals.
MR Imaging Protocol
All sites used 1.5-T units (Signa; GE Medical Systems). MR imaging was performed with a multicoil array or a built-in body coil. Whenever possible, the multicoil array was used for pelvic imaging with the body coil reserved for abdominal imaging. However, the body coil alone was used for imaging the pelvis and abdomen in the following situations: a mass larger than 15 cm in diameter or a patient with severe obesity or ascites. All patients were requested to fast for at least 3 hours before MR imaging and received 1 mg of intramuscularly administered glucagon (unless contraindicated) before MR imaging.
The pelvis was imaged with an axial fast spin-echo T2-weighted sequence (5,000–6,000/102–126 [repetition time msec/echo time msec]) with an echo train length of 16, a 5–10-mm section thickness, and a 0–2.5-mm intersection gap. The matrix size was 256 x 256 with two signals acquired. This sequence was repeated in the coronal and sagittal planes as indicated. An axial T1-weighted spin-echo sequence (600–800/11–20) with spatial resolution similar to that of the T2-weighted sequence was then performed. The T1-weighted sequence was repeated with fat suppression after intravenous injection of 10–20 mL of gadolinium chelate.
The remainder of the abdomen and pelvis was imaged with an axial T2-weighted fast spin-echo sequence and an axial fat-suppressed T1-weighted spin-echo sequence. The T2-weighted sequence (4,000/102–126) had an echo train length of 16, an 8–10-mm section thickness, and a 1.0–2.5-mm gap. The matrix was 256 x 192 or 256 with two to four signals acquired. This sequence was repeated in the coronal and sagittal planes as indicated. The T1-weighted sequence (400–600/11–20) had an 8–10-mm section thickness with a 2-mm gap. The matrix was 256 x 128–192 with two signals acquired.
Quality Control
Uniform standards were provided for all institutions and all imagers at the onset of the study and at routine intervals throughout the study. These standards covered hard-copy and recorded assessment; test objects, distance, resolution, and penetration for US; test patterns and low-contrast detection for CT; and phantoms and surface coil assessment for MR imaging.
Interpretation of Studies
Three radiologists (one for each modality) from each institution read the protocol imaging studies. They read studies only from their own institution. The readers had access to all clinical information, available laboratory test results, and admitting imaging studies; they had no knowledge of the results of the other protocol imaging studies or of surgery and histopathologic analysis.
In the analysis of the ovary and adnexa, Doppler US was used to evaluate the arterial flow of the detected complex and solid components of one or two masses. Up to three spectral arterial waveform samples were taken from thick walls, thick septa, and solid components (when present). If possible, three waveform samples were also obtained from both ovaries (normal or abnormal). For each waveform, a pulsatility index and resistive index were calculated; thresholds for malignancy were a pulsatility index of less than 1.0 and a resistive index of less than 0.4 (10,11). To increase the likelihood of diagnosing malignancy, only the lowest values of these indexes were used. Each waveform was also analyzed for the presence of a diastolic notch, a finding thought by some to be diagnostic of benign lesions. Analysis of the ovaries and adnexa with CT and MR imaging involved measurement and characterization of one or two complex and/or solid masses for all components ranging from cystic to solid, including identification of necrosis and fat. Both ovaries (normal or abnormal) were also analyzed.
Malignant spread into the remainder of the pelvis and into the abdomen was defined as complex and/or solid masses and/or complex fluid on US, CT, and MR images. In the pelvis, all organs and regions were evaluated including the fallopian tubes, uterus, parametrium, cul-de-sac, pelvic side walls, vagina, and lymph nodes. In the abdomen, all organs; the bowel; the mesenteric, omental, and peritoneal regions; and the lymph nodes were evaluated. Complex ascites in both the pelvis and abdomen and pleural effusions were noted. A total of 59 sites were evaluated including five peritoneal, three mesenteric, nine intestinal, and eight nodal regions.
The presence or absence of every finding was evaluated with the following five-point scale: 0 = normal, 1 = probably normal, 2 = indeterminate, 3 = probably abnormal, and 4 = abnormal. Each radiologist then determined a final stage based on the modified International Federation of Gynecology and Obstetrics staging classification (31) (Table 1).
|
Pathologists from each institution performed the interpretation in a routine manner using the revised World Health Organization histologic classification for ovarian neoplasms (32,33). For the primary ovarian tumor, a minimum of one block per centimeter of the greatest tissue diameter was examined along with additional specimens if the tumor extended to the surface. If there was ambiguity about the histologic type, another pathologist made the final decision.
The standard of reference was a combination of surgical and histopathologic results, which were classified according to the modified International Federation of Gynecology and Obstetrics staging classification (Table 1). Although there are different types of ovarian malignancies, all are similarly staged. Patients who had nonovarian cancer with spread to the ovaries were evaluated as if they had primary ovarian cancer. For example, a primary malignancy of the appendix with spread to the ovaries was classified as stage III disease because ovarian cancer with appendiceal spread would have been classified as stage III ovarian cancer.
Quality Review Committee
The quality review committee consisted of four radiologists (A.B.K., P.H.A., D.G.M., E.S.S.) from two institutions; two of the radiologists had expertise in US and CT, and two had expertise in MR imaging. They assessed the studies for quality and adherence to study protocols (without reinterpreting the studies) in their areas of expertise no later than 3 months after the start of patient accrual and at regular intervals thereafter.
Statistical Methods
The abilities of the three modalities to allow discrimination between benign cases and those with cancer in any anatomic region of the pelvis and abdomen were assessed. As previously stated, Doppler US, CT, and MR imaging were used to evaluate the ovaries and adnexa for malignancy; conventional US, CT, and MR imaging were used to evaluate the extraovarian pelvis and abdomen for spread.
For each institution, the accuracies of the modalities were measured by estimating the area under the receiver operating characteristic (ROC) curve, and homogeneity among the five institution-specific areas under the ROC curve was tested with the 
2 method. Modality-specific areas under the ROC curve were estimated by pooling the institution-specific area estimates by means of weighted averages. Pairwise comparisons between modalities were performed for each institution by testing the difference between areas under the ROC curve obtained for each modality pair. Thus, whereas overall estimated areas were determined for each modality across institutions, comparisons between modalities were made by comparing results in the subset of patients who underwent both modalities. Modality-specific differences were addressed by pooling these differences across institutions. Estimation and comparison of the areas under the ROC curve were based on the nonparametric approach of DeLong et al (34).
The accuracy of each modality in detection of malignancy was estimated separately for the three anatomic regions (ovaries and adnexa, extraovarian pelvis, abdomen and beyond) by computing the relevant areas under the ROC curve. For detection of malignancy in the ovaries and adnexa, the accuracies of Doppler US, CT, and MR imaging were compared. For detection of spread into the extraovarian pelvis and abdomen and beyond, the accuracies of conventional US, CT, and MR imaging were compared. For the ovaries and adnexa, the accuracy of detection was based on the presence or absence of malignancy in the ovaries and adnexa. For the extraovarian pelvis, the accuracy of detection was based on the presence or absence of extraovarian pelvic disease without consideration of the presence or absence of disease in the ovaries and adnexa or in the abdomen. For the abdomen and beyond, the accuracy of detection was based on the presence or absence of disease in this area without consideration of the presence or absence of disease in the pelvis. Analysis of each specific pelvic and abdominal site, however, was beyond the scope of the study.
For all ROC curve analyses, modality-specific parametric ROC curves were obtained by fitting an ordinal regression model, accounting for institution-specific cutoff points on the latent variable scale (35). Both empiric institution-specific ROC curves and smooth parametric curves were generated.
Agreement between the results of imaging and the standard of reference (surgical and histopathologic analysis) was based on the reader's final assessment and was measured with 
statistics. The stages of disease were grouped as follows in a 3 x 3 table: benign lesions and stage I cancer, stage II cancer, and stage III or IV cancer. Benign lesions and stage I cancer were grouped to assess the abilities of the imaging modalities to allow distinction between disease confined to the ovaries and spread of cancer (stage II to IV). Cases of nonadvanced disease (benign lesions, stage I and II cancer) were grouped to assess the abilities of the imag-ing modalities to demonstrate abdominal spread (stage III and IV) because the surgical approach is different. This issue was addressed by using the radiologist's final assessment of disease stage to estimate and compare modality-specific sensitivities and specificities by means of the McNemar test for correlated proportions.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Of the 280 patients, 118 (42.1%) had malignancies (Table 2): 87 primary ovarian, 27 metastatic (nonovarian) with spread to the ovaries, and four metastatic but with concomitant benign ovarian disease. The cancers were staged as follows: 35 confined to the ovaries (stage I), 13 involving the ovaries and adjacent pelvic structures (stage II), and 70 with abdominal spread (stage III or IV).
|
2 test]). According to the inclusion criteria, all patients had to be enrolled in at least two modalities. However, not all patients completed their studies due to noncompliance or physical disability, which was often caused by advanced disease. The results of all completed studies were included in the evaluation. All three modalities were performed in 103 patients (36.8%). Only two modalities were performed in 170 patients (60.7%): Doppler and conventional US and CT in 97, Doppler and conventional US and MR imaging in 60, and CT and MR imaging in 13. Only one modality was performed in seven patients (2.5%): Doppler and conventional US in four and MR imaging in three. In summary, the modality accuracy estimates and the statistical analyses were based on the results of 264 Doppler and conventional US examinations, 213 CT examinations, and 179 MR imaging examinations.
In discrimination between cases of benign and cases of malignant disease in any region of the pelvis and abdomen, all three modalities had approximately equal overall pooled areas under the ROC curve of 0.91 (95% CI = 0.87, 0.95) (Table 3, Fig 1). There was a statistically significant difference between institutions only for MR imaging, with the institution-specific areas ranging from 0.77 to 0.96 (P = .045) because of a considerably smaller area for institution E. Although the range for CT was larger, from 0.74 to 0.97, no statistically significant difference between institutions was detected.
|
|
|
|
|
|
|
|
|
The statistics for agreement between the results of a modality (based on the readers' final evaluation) and the standard of reference were similar for staging with US, CT, and MR imaging: 0.66 for US (95% CI = 0.56, 0.74); 0.65 for CT (95% CI = 0.55, 0.75); and 0.70 for MR imaging (95% CI = 0.60, 0.81).
In the differentiation of nonadvanced disease (benign lesions, stage I and II cancers) from advanced malignancy (stage III and IV cancers) with US, eight of 197 cases of nonadvanced disease were overstaged as advanced malignancy (false-positive rate = 0.041) and 17 of 67 cases of advanced malignancy were understaged as stage I or II (false-negative rate = 0.254) (Table 6). With CT, 17 of 161 cases of nonadvanced disease were overstaged as advanced malignancy (false-positive rate = 0.106) and four of 50 cases of advanced malignancy were understaged as stage I or II (false-negative rate = 0.080). With MR imaging, 17 of 139 cases of nonadvanced disease were overstaged as advanced malignancy (false-positive rate = 0.122) and one of 41 cases of advanced malignancy was understaged as stage I or II (false-negative rate = 0.024). Five cases were overstaged with both CT and MR imaging. In summary, the specificity was 96% for US, 89% for CT, and 88% for MR imaging and the sensitivity was 75% for US, 92% for CT, and 98% for MR imaging. On the basis of the common cases for each modality pair, the following differences were statistically significant: (a) the difference in specificity between US and MR imaging (P = .018), with US more specific; (b) the difference in sensitivity between US and CT (P = .014), with CT more sensitive; and (c) the difference in sensitivity between US and MR imaging (P = .003), with MR imaging more sensitive.
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The percentage of stage III and IV cancers was high, almost 60% (70 of 118 cases). However, the malignancy rate of 42.1% was more than double those in previous studies (36,37). This result was likely due to the referral source, which was specialists in gynecologic oncology.
There was remarkably little variability among participating institutions and modalities. Recruitment patterns were similar with no statistically significant differences in the percentages of ovarian cancers and their stages (Table 2). This fact suggests that the results are reproducible and can likely be generalized to other centers.
The three imaging modalities yielded similar estimated areas under the ROC curve for discrimination between benign disease and cancer in all regions, with a mean area of 0.91 for all three modalities (Table 3). Although differentiation of benign from malignant disease is obviously clinically important and these detection rates are higher than those previously reported, they are likely still not high enough for surgery to be avoided in most cases.
For the ovaries and adnexa, the intent of this study was not to evaluate the efficacy of screening or mass detection. Rather, because all of the patients had a mass detected at US, the intent was to evaluate the abilities of Doppler US, CT, and MR imaging to allow diagnosis of malignancy in the mass. For this application, the estimated areas under the ROC curve for each modality revealed that MR imaging and CT were superior to Doppler US (Table 4). It is unfortunate (but unavoidable) that our study design did not allow statistical evaluation of the performances of CT and MR imaging versus that of combined Doppler and conventional US (38).
The relatively poor performance of Doppler US has been discussed without identification of a unifying cause (13–15,39). The neovascular response of ovarian cancer may be incomplete or nonspecific; thus, when intratumoral arterial flow is detected with Doppler US, differentiation of benign from malignant processes is often not possible. It has been suggested that central flow, increased impedance, elevated peak systolic velocity, loss of the diastolic notch, and low Doppler US indexes favor malignancy. None of these have shown a high enough diagnostic accuracy. Therefore, whatever the reason, the present lack of specificity makes Doppler US a poor diagnostic tool. It is possible that different numeric cutoff values for the Doppler indexes (40–42) may provide more encouraging results, although the considerable overlap of the indexes in benign and malignant tumors makes this possibility unlikely. Perhaps use of intravenous contrast agents to enhance the tumor and its vascularity may improve diagnostic accuracy (43,44).
Areas under the ROC curve reflect the tendencies of readers to overstage and understage disease. However, these areas do not reveal differences in overstaging and understaging at the routine interpretative cutoffs used by most readers. Thus, we also analyzed the data according to the way they are actually received and used by the referring clinicians (Table 6). Although trends were difficult to discern because the "off-diagonal" numbers were small (Table 6), the data suggest the following: (a) For pelvic spread (stage II cancers), there was considerable understaging (false-negative cases) and overstaging (false-positive cases) with all three modalities. (b) For differentiation of pelvic malignancy (stage I or II) from abdominal spread (stage III or IV), abdominal spread was more likely to be understaged with US and pelvic malignancy was more likely to be overstaged with CT and MR imaging.
There are problems with both overstaging and understaging. Overstaging leads to a wider surgical approach but makes it more likely that malignancy will not be overlooked; understaging causes extension of the originally planned pelvic surgery with the potential for tumor to be overlooked. The extent of surgical expertise devoted to the case may also be affected. In most centers, the greater the suspicion of abdominal spread, the more likely it is that a specialist in gynecologic oncology will perform the exploration. This point is important because surgery by specialists leads to better staging and maximal debulking (cytoreduction) and to longer survival (45–48). Although one could argue that it is always better to overstage so that malignancy is not overlooked, it is more important to accurately distinguish malignancy that has spread into the abdomen (stage III or IV) from malignancy confined to the pelvis (stage I or II).
Our results show remarkable accuracy in demonstrating the extent of malignant spread for all three imaging modalities. Although US, CT, and MR imaging may have demonstrated disease in different sites (eg, the mesentery rather than paracolic gutters), all three modalities had high staging accuracies at ROC curve analysis of 0.91. It is therefore difficult to suggest a simple algorithm for evaluation of women with ovarian masses. However, US is often the initial imaging study in evaluation of a suspected ovarian abnormality. If a complex or solid ovarian mass and abdominal spread are detected with US, our data support the accuracy of a diagnosis of stage III disease. Similarly, CT and MR imaging allow accurate diagnosis of a complex or solid ovarian mass with stage III spread. Because of the importance of not understaging abdominal malignancy as disease limited to the pelvis, if stage III cancer is not detected at initial abdominal US, CT or MR imaging should be performed because of their higher sensitivities in staging. Whatever the modality used, it is hoped that correct staging of advanced disease will lead to appropriate referral to a specialist in gynecologic oncology.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
3 Current address: Dept of Radiology, University of Wisconsin Medical School, Madison.
Author contributions: Guarantor of integrity of entire study, A.B.K.; study concepts, A.B.K.; study design, A.B.K., J.V.T., B.J.M., C.M.C.T., U.M.H., R.L.B., P.H.A., D.J.C.; definition of intellectual content, A.B.K., C.M.C.T.; literature research, J.V.T., R.J.W., A.B.K., C.M.C.T., I.R.F.; clinical studies, A.B.K., C.M.C.T., U.M.H., P.H.A., R.L.B., R.J.W., I.R.F., J.E.K., E.S.S., D.G.M., S.G.S., D.L.B., S.S., B.G.C., J.H.E., R.J.K.; data acquisition and analysis, J.V.T., B.J.M., D.J.C.; statistical analysis, J.V.T., B.J.M., D.J.C.; manuscript preparation, all authors; manuscript editing, A.B.K., B.J.M., J.V.T., C.M.C.T.; manuscript review, A.B.K., C.M.C.T.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  N. A. Telischak, B. M. Yeh, B. N. Joe, A. C. Westphalen, L. Poder, and F. V. Coakley MRI of Adnexal Masses in Pregnancy Am. J. Roentgenol., August 1, 2008; 191(2): 364 - 370. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S Ghattamaneni, M J Weston, and J A Spencer Imaging in endometriosis Imaging, December 1, 2007; 19(4): 345 - 368. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. D. Kirkpatrick, M. A. Brewer, and U. Utzinger Endogenous Optical Biomarkers of Ovarian Cancer Evaluated with Multiphoton Microscopy Cancer Epidemiol. Biomarkers Prev., October 1, 2007; 16(10): 2048 - 2057. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Barua, J. S. Abramowicz, J. M. Bahr, P. Bitterman, A. Dirks, K. A. Holub, E. Sheiner, M. J. Bradaric, S. L. Edassery, and J. L. Luborsky Detection of Ovarian Tumors in Chicken by Sonography: A Step Toward Early Diagnosis in Humans? J. Ultrasound Med., July 1, 2007; 26(7): 909 - 919. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M V Pakkal and M Balogun Imaging of ovarian cancer Imaging, August 1, 2006; 18(1): 20 - 27. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W.-C. Chang, M. D. Meux, B. M. Yeh, A. Qayyum, B. N. Joe, L.-m. Chen, and F. V. Coakley CT and MRI of Adnexal Masses in Patients with Primary Nonovarian Malignancy. Am. J. Roentgenol., April 1, 2006; 186(4): 1039 - 1045. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J A Spencer A multidisciplinary approach to ovarian cancer at diagnosis Br. J. Radiol., October 1, 2005; 78(Special_Issue_2): S94 - S102. [Full Text] [PDF]
|
|
|
|
 |
|
 K. Kinkel, Y. Lu, A. Mehdizade, M.-F. Pelte, and H. Hricak Indeterminate Ovarian Mass at US: Incremental Value of Second Imaging Test for Characterization--Meta-Analysis and Bayesian Analysis Radiology, July 1, 2005; 236(1): 85 - 94. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. R. Aberle, C. Chiles, C. Gatsonis, B. J. Hillman, C. D. Johnson, B. L. McClennan, D. G. Mitchell, E. D. Pisano, M. D. Schnall, and A. G. Sorensen Imaging and Cancer: Research Strategy of the American College of Radiology Imaging Network Radiology, June 1, 2005; 235(3): 741 - 751. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. N. Low, B. Duggan, R. M. Barone, F. Saleh, and S. Y. T. Song Treated Ovarian Cancer: MR Imaging, Laparotomy Reassessment, and Serum CA-125 Values Compared with Clinical Outcome at 1 Year Radiology, June 1, 2005; 235(3): 918 - 926. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Goldberg-Zimring, B. Shalmon, K. H. Zou, H. Azhari, D. Nass, and A. Achiron Assessment of Multiple Sclerosis Lesions with Spherical Harmonics: Comparison of MR Imaging and Pathologic Findings Radiology, June 1, 2005; 235(3): 1036 - 1044. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. T. Beinfeld, E. Wittenberg, and G. S. Gazelle Cost-effectiveness of Whole-Body CT Screening Radiology, February 1, 2005; 234(2): 415 - 422. [Abstract] [Full Text] [PDF]
|
|
|
