-
[show abstract]
[hide abstract]
ABSTRACT: AIM: To compare the diagnostic accuracy of subjective ultrasound assessment to objective measurement techniques in the evaluation of myometrial and cervical invasion in women with endometrial cancer. METHODS: Prospective multicenter study including 144 women with endometrial cancer undergoing transvaginal ultrasound. Myometrial and cervical invasion was subjectively evaluated, and objectively measured in different ways: ;endometrial thickness, tumor/uterine anterio-posterior (AP) ratio, minimal tumor free margin, minimal tumor free margin/uterine AP diameter ratio, tumor volume (3D), tumor /uterine volume (3D) ratio, distance from outer cervical os to the lower margin of tumor (Dist-OCO). Histological assessment from hysterectomy was golden standard. RESULTS: The sensitivity (79%) and specificity (76%) of tumor/uterine AP diameter (at cut-off 0.53) was not significantly different from subjective evaluation (sensitivity 77%, p = 0.44, specificity 81%, p = 0.32), all other objective measurement techniques had either a significantly lower sensitivity or a lower specificity. Fixing sensitivity at the same level as subjective evaluation all objective measurement techniques, except minimal tumor free margin/uterine AP diameter ratio, had a significantly lower specificity. Dist-OCO was the only parameter that might have potential to predict cervical invasion, it had a non-significantly higher sensitivity than subjective evaluation (vs. 73%, vs. 54% p = 0.06), but a significantly lower specificity (63% vs. 93% p < 0.001). CONCLUSION: Subjective assessment of cervical and myometrial invasion is as good as or better than any objective measurement technique. The tumor/uterine AP ratio and minimal tumor free margin/uterine AP diameter ratio seem to be the best objective measurement techniques to predict deep myometrial invasion. It remains to be shown if objective measurements are useful to predict cervical invasion.
Ultrasound in Obstetrics and Gynecology 05/2013; · 3.01 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: OBJECTIVE: The identification of novel biomarkers led to the development of the ROMA algorithm incorporating both HE4 and CA125 to predict malignancy in women with a pelvic mass. An ultrasound based prediction model (LR2) developed by the International Ovarian Tumor Analysis (IOTA) study offers better diagnostic performance than CA125 alone. In this study we compared the diagnostic accuracy between LR2 and ROMA. METHODS: This study included women with a pelvic mass scheduled for surgery and enrolled in a previous prospective diagnostic accuracy study. Experienced ultrasound examiners, general gynecologists and trainees supervised by one of the experts performed the preoperative transvaginal ultrasound examinations. Serum biomarkers were taken prior to surgery. Accuracy of LR2 and ROMA was estimated at completion of this study and did not form part of the decision making process. Final outcome was histology of removed tissues and surgical stage if relevant. RESULTS: In total 360 women were evaluated. 216 women had benign disease and 144 a malignancy. Overall test performance of LR2 (AUC 0.952) with 94% sensitivity and 82% specificity was significantly better than ROMA (AUC 0.893) with 84% sensitivity and 80% specificity. Difference in AUC was 0.059 (95% CI: 0.026-0.091; P-value 0.0004). Similar results were obtained when stratified for menopausal status. CONCLUSION: LR2 shows better diagnostic performance than ROMA for the characterization of a pelvic mass in both pre- and postmenopausal women. These findings suggest that HE4 and CA125 may not play an important role in the diagnosis of ovarian cancer if good quality ultrasonography is available.
Gynecologic Oncology 01/2013; · 3.89 Impact Factor
-
Lieveke Ameye,
Dirk Timmerman,
Lil Valentin,
Dario Paladini,
Jingzhang, Caroline Van Holsbeke,
Andrea A Lissoni,
Luca Savelli,
Joan Veldman,
Antonia C Testa,
Frederic Amant,
Sabine Van Huffel,
Tom Bourne
[show abstract]
[hide abstract]
ABSTRACT: Objective If we exclude masses likely to be judged as easy and "instant" to diagnose by an ultrasound examiner, what is the diagnostic performance of ultrasound based simple rules, RMI, two logistic regression models and real-time subjective assessment by experienced ultrasound examiners? Patients and methods 3511 patients with at least one persistent adnexal mass preoperatively underwent transvaginal ultrasonography to assess tumor morphology and vascularity. They were included in two consecutive prospective studies by the International Ovarian Tumor Analysis (IOTA) group: phase 1 (1999-2005) development of the simple rules and logistic regression models LR1 and LR2, and phase 2 a validation study (2005-2007). Results Almost half of the cases (43%) were identified as "instant" to diagnose on the basis of descriptors applied to the database. To assess the performance in the more difficult "non-instant" masses, we used only phase 2 data (N = 1036). The sensitivity of LR2 was 88%, RMI 41% and subjective assessment 87%. The specificity of LR2 was 67%, RMI 90% and subjective assessment 86%. The simple rules yielded a conclusive result in almost 2/3 of the masses where they resulted in sensitivity and specificity similar to real-time subjective assessment by experienced ultrasound examiners: sensitivity 89% vs. 89% (P-value 0.76), specificity 91% vs. 91% (P-value 0.65). If we apply a three-step strategy with easy "instant" diagnoses as step 1, simple rules as step 2 and subjective assessment by an experienced ultrasound examiner in the remaining masses as step 3 we obtain a sensitivity of 92% and specificity of 92% compared to sensitivity 90% (P = 0.03) and specificity 93% (P = 0.44) when using real-time subjective assessment by experts in all tumors, respectively. Conclusion A diagnostic strategy using simple descriptors and ultrasound rules when applied to the variables contained in the IOTA database obtains results that are at least as good as those obtained by subjective assessment of a mass by an expert. Copyright © 2012 ISUOG. Published by John Wiley & Sons, Ltd.
Ultrasound in Obstetrics and Gynecology 04/2012; · 3.01 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Objectives: To describe clinical and ultrasound features of Brenner tumors of the ovary. Methods: In this retrospective study, the databases of the International Ovarian Tumor Analysis studies and one tertiary center were searched to identify patients who had undergone an ultrasound scan before surgery for an adnexal mass that proved to be a Brenner tumor. Twenty-eight patients with 29 Brenner tumors were included. Most tumors had been collected within the frame of the IOTA studies. An experienced ultrasound examiner reviewed available ultrasound images (available for 14 tumors) searching for a specific pattern of Brenner tumors. Results: Most patients were postmenopausal and asymptomatic. Twenty-four (83%) tumors were benign, two (7%) were borderline and three (10%) malignant. Most benign Brenner tumors (17/24, 71%) contained solid components and manifested no or minimal blood flow at Doppler examination (19/24, 79%). Information about calcifications was available for 15 benign Brenner tumors and in 13 (87%) calcifications were present. The five borderline and invasively malignant Brenner tumors contained solid components less often than the benign ones (3/5, 60%) and were more richly vascularized at Doppler examination. Information about calcifications was available for four malignant Brenner tumors and in three (75%) calcifications were present. Conclusion: We failed to demonstrate ultrasound features specific for Brenner tumors. A prospective study is needed to determine if ultrasound features of calcifications can discriminate between Brenner tumors and other types of ovarian tumor. Copyright © 2012 ISUOG. Published by John Wiley & Sons, Ltd.
Ultrasound in Obstetrics and Gynecology 03/2012; · 3.01 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Despite the rise of high-throughput technologies, clinical data such as age, gender and medical history guide clinical management for most diseases and examinations. To improve clinical management, available patient information should be fully exploited. This requires appropriate modeling of relevant parameters.
When kernel methods are used, traditional kernel functions such as the linear kernel are often applied to the set of clinical parameters. These kernel functions, however, have their disadvantages due to the specific characteristics of clinical data, being a mix of variable types with each variable its own range. We propose a new kernel function specifically adapted to the characteristics of clinical data.
The clinical kernel function provides a better representation of patients' similarity by equalizing the influence of all variables and taking into account the range r of the variables. Moreover, it is robust with respect to changes in r. Incorporated in a least squares support vector machine, the new kernel function results in significantly improved diagnosis, prognosis and prediction of therapy response. This is illustrated on four clinical data sets within gynecology, with an average increase in test area under the ROC curve (AUC) of 0.023, 0.021, 0.122 and 0.019, respectively. Moreover, when combining clinical parameters and expression data in three case studies on breast cancer, results improved overall with use of the new kernel function and when considering both data types in a weighted fashion, with a larger weight assigned to the clinical parameters. The increase in AUC with respect to a standard kernel function and/or unweighted data combination was maximum 0.127, 0.042 and 0.118 for the three case studies.
For clinical data consisting of variables of different types, the proposed kernel function--which takes into account the type and range of each variable--has shown to be a better alternative for linear and non-linear classification problems.
Artificial intelligence in medicine 11/2011; 54(2):103-14. · 1.65 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Hereditary hemorrhagic telangiectasia (HHT) is a rare but life-threatening disease characterized by multi system telangiectasias and arteriovenous malformations (AVM). Complications in adults have been reported extensively, but neonatal (NN) complications have only been published in incidental case reports. In this paper, we present a literature review on NN pulmonary AVM related to HHT, following our own experience with a NN death due to this disease. As prenatal diagnosis of pulmonary AVM is feasible, we recommend that a family history of HHT should be an indication for expertise prenatal anomaly scanning, in order to organise optimal NN support at birth.
The journal of maternal-fetal & neonatal medicine: the official journal of the European Association of Perinatal Medicine, the Federation of Asia and Oceania Perinatal Societies, the International Society of Perinatal Obstetricians 11/2011; 25(8):1494-8. · 1.36 Impact Factor
-
E Epstein, C Van Holsbeke,
F Mascilini,
A Måsbäck,
P Kannisto,
L Ameye,
D Fischerova,
G Zannoni,
V Vellone,
D Timmerman,
A C Testa
[show abstract]
[hide abstract]
ABSTRACT: To describe the gray-scale and vascular characteristics of endometrial cancer in relation to stage, grade and size using two-dimensional (2D)/three-dimensional (3D) transvaginal ultrasound.
This was a prospective multicenter study including 144 women with endometrial cancer undergoing transvaginal ultrasound before surgery. The sonographic characteristics assessed were echogenicity, endometrial/myometrial border, fibroids, vascular pattern, color score and tumor/uterus anteroposterior (AP) ratio. Histological assessment of tumor stage, grade, type and growth pattern was performed.
Hyperechoic or isoechoic tumors were more often seen in Stage IA cancer, whereas mixed or hypoechoic tumors were more often found in cancers of Stage IB or greater (P = 0.003). Hyperechogenicity was more common in Grade 1-2 tumors (i.e. well or moderately differentiated) (P = 0.02) and in tumors with a tumor/uterine AP ratio of < 50% (P = 0.002), whereas a non-hyperechoic appearance was more commonly found in Grade 3 tumors (i.e. poorly differentiated) and in tumors with a tumor/uterine AP ratio of ≥ 50%. Multiple global vessels were more often seen in tumors of Stage IB or greater than in Stage IA tumors (P = 0.02), in Grade 3 tumors than in Grade 1 and 2 tumors (P = 0.02) and in tumors with a tumor/uterine AP ratio of ≥ 50% (P < 0.001). A moderate/high color score was significantly more common in tumors of higher stage (P = 0.03) and larger size (P = 0.001).
The sonographic appearance of endometrial cancer is significantly associated with tumor stage, grade and size. More advanced tumors often have a mixed/hypoechoic echogenicity, a higher color score and multiple globally entering vessels, whereas less advanced tumors are more often hyperechoic and have no or a low color score.
Ultrasound in Obstetrics and Gynecology 05/2011; 38(5):586-93. · 3.01 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Hitherto, risk prediction models for preoperative ultrasound-based diagnosis of ovarian tumors were dichotomous (benign versus malignant). We develop and validate polytomous models (models that predict more than two events) to diagnose ovarian tumors as benign, borderline, primary invasive or metastatic invasive. The main focus is on how different types of models perform and compare.
A multi-center dataset containing 1066 women was used for model development and internal validation, whilst another multi-center dataset of 1938 women was used for temporal and external validation. Models were based on standard logistic regression and on penalized kernel-based algorithms (least squares support vector machines and kernel logistic regression). We used true polytomous models as well as combinations of dichotomous models based on the 'pairwise coupling' technique to produce polytomous risk estimates. Careful variable selection was performed, based largely on cross-validated c-index estimates. Model performance was assessed with the dichotomous c-index (i.e. the area under the ROC curve) and a polytomous extension, and with calibration graphs.
For all models, between 9 and 11 predictors were selected. Internal validation was successful with polytomous c-indexes between 0.64 and 0.69. For the best model dichotomous c-indexes were between 0.73 (primary invasive vs metastatic) and 0.96 (borderline vs metastatic). On temporal and external validation, overall discrimination performance was good with polytomous c-indexes between 0.57 and 0.64. However, discrimination between primary and metastatic invasive tumors decreased to near random levels. Standard logistic regression performed well in comparison with advanced algorithms, and combining dichotomous models performed well in comparison with true polytomous models. The best model was a combination of dichotomous logistic regression models. This model is available online.
We have developed models that successfully discriminate between benign, borderline, and invasive ovarian tumors. Methodologically, the combination of dichotomous models was an interesting approach to tackle the polytomous problem. Standard logistic regression models were not outperformed by regularized kernel-based alternatives, a finding to which the careful variable selection procedure will have contributed. The random discrimination between primary and metastatic invasive tumors on temporal/external validation demonstrated once more the necessity of validation studies.
BMC Medical Research Methodology 10/2010; 10:96. · 2.67 Impact Factor
-
D Timmerman,
B Van Calster,
A C Testa,
S Guerriero,
D Fischerova,
A A Lissoni, C Van Holsbeke,
R Fruscio,
A Czekierdowski,
D Jurkovic,
L Savelli,
I Vergote,
T Bourne,
S Van Huffel,
L Valentin
[show abstract]
[hide abstract]
ABSTRACT: The aims of the study were to temporally and externally validate the diagnostic performance of two logistic regression models containing clinical and ultrasound variables in order to estimate the risk of malignancy in adnexal masses, and to compare the results with the subjective interpretation of ultrasound findings carried out by an experienced ultrasound examiner ('subjective assessment').
Patients with adnexal masses, who were put forward by the 19 centers participating in the study, underwent a standardized transvaginal ultrasound examination by a gynecologist or a radiologist specialized in ultrasonography. The examiner prospectively collected information on clinical and ultrasound variables, and classified each mass as benign or malignant on the basis of subjective evaluation of ultrasound findings. The gold standard was the histology of the mass with local clinicians deciding whether to operate on the basis of ultrasound results and the clinical picture. The models' ability to discriminate between malignant and benign masses was assessed, together with the accuracy of the risk estimates.
Of the 1938 patients included in the study, 1396 had benign, 373 had primary invasive, 111 had borderline malignant and 58 had metastatic tumors. On external validation (997 patients from 12 centers), the area under the receiver-operating characteristics curve (AUC) for a model containing 12 predictors (LR1) was 0.956, for a reduced model with six predictors (LR2) was 0.949 and for subjective assessment was 0.949. Subjective assessment gave a positive likelihood ratio of 11.0 and a negative likelihood ratio of 0.14. The corresponding likelihood ratios for a previously derived probability threshold (0.1) were 6.84 and 0.09 for LR1, and 6.36 and 0.10 for LR2. On temporal validation (941 patients from seven centers), the AUCs were 0.945 (LR1), 0.918 (LR2) and 0.959 (subjective assessment).
Both models provide excellent discrimination between benign and malignant masses. Because the models provide an objective and reasonably accurate risk estimation, they may improve the management of women with suspected ovarian pathology.
Ultrasound in Obstetrics and Gynecology 03/2010; 36(2):226-34. · 3.01 Impact Factor
-
C Van Holsbeke,
V Van Belle,
F P G Leone,
S Guerriero,
D Paladini,
G B Melis,
S Greggi,
D Fischerova,
E De Jonge,
P Neven,
T Bourne,
L Valentin,
S Van Huffel,
D Timmerman
[show abstract]
[hide abstract]
ABSTRACT: To determine the sensitivity and specificity of the 'ovarian crescent sign' (OCS)-a rim of normal ovarian tissue seen adjacent to an ipsilateral adnexal mass-as a sonographic feature to discriminate between benign and malignant adnexal masses.
The patients included were a subgroup of patients participating in the International Ovarian Tumor Analysis (IOTA) Phase 2 study, which is an international multicenter study. The subgroup comprised 1938 patients, with an adnexal mass, recruited from 19 ultrasound centers in different countries. All patients were scanned using the same standardized ultrasound protocol. Information on more than 40 demographic and ultrasound variables were collected, but the evaluation of the OCS was optional. Only patients from centers that had evaluated the OCS in > or = 90% of their cases were included. The gold standard was the histological diagnosis of the adnexal mass. The ability of the OCS to discriminate between borderline or invasively malignant vs. benign adnexal masses, as well as between invasively malignant vs. other (benign and borderline) tumors, was determined and compared with the performance of subjective evaluation of ultrasound findings by the ultrasound examiner.
The OCS was evaluated in 1377 adnexal masses from 12 centers, 938 (68%) masses being benign, 86 (6%) borderline, 305 (22%) primary invasive and 48 (3%) metastases. The OCS was present in 398 (42%) of 938 benign masses, in 14 (16%) of 86 borderline tumors, in 18 (6%) of 305 primary invasive tumors (one malignant struma ovarii, one uterine clear cell adenocarcinoma and 16 epithelial carcinomas, i.e. four Stage I and 12 Stage II-IV) and in two (4%) of 48 ovarian metastases. Hence, the sensitivity and specificity for absent OCS to identify a malignancy was 92% and 42%, respectively, and the positive and negative likelihood ratios (LR+ and LR-, respectively) were 1.60 and 0.18. Subjective impression performed significantly better than the OCS. Sensitivity and specificity were 90% and 92%, respectively, LR+ was 11.0 and LR- was 0.10. For discrimination between invasive vs. benign or borderline tumors, the sensitivity for absent OCS was 94%, the specificity was 40%, the LR+ was 1.58 and the LR- was 0.14.
This study confirms previous reports that the presence of the OCS decreases the likelihood of invasive malignancy in adnexal masses. However it is a poor discriminator between benign and malignant adnexal masses.
Ultrasound in Obstetrics and Gynecology 03/2010; 36(1):81-7. · 3.01 Impact Factor
-
Dirk Timmerman,
Lieveke Ameye,
Daniela Fischerova,
Elisabeth Epstein,
Gian Benedetto Melis,
Stefano Guerriero, Caroline Van Holsbeke,
Luca Savelli,
Robert Fruscio,
Andrea Alberto Lissoni,
Antonia Carla Testa,
Joan Veldman,
Ignace Vergote,
Sabine Van Huffel,
Tom Bourne,
Lil Valentin
[show abstract]
[hide abstract]
ABSTRACT: To prospectively assess the diagnostic performance of simple ultrasound rules to predict benignity/malignancy in an adnexal mass and to test the performance of the risk of malignancy index, two logistic regression models, and subjective assessment of ultrasonic findings by an experienced ultrasound examiner in adnexal masses for which the simple rules yield an inconclusive result.
Prospective temporal and external validation of simple ultrasound rules to distinguish benign from malignant adnexal masses. The rules comprised five ultrasonic features (including shape, size, solidity, and results of colour Doppler examination) to predict a malignant tumour (M features) and five to predict a benign tumour (B features). If one or more M features were present in the absence of a B feature, the mass was classified as malignant. If one or more B features were present in the absence of an M feature, it was classified as benign. If both M features and B features were present, or if none of the features was present, the simple rules were inconclusive.
19 ultrasound centres in eight countries.
1938 women with an adnexal mass examined with ultrasound by the principal investigator at each centre with a standardised research protocol. Reference standard Histological classification of the excised adnexal mass as benign or malignant.
Diagnostic sensitivity and specificity.
Of the 1938 patients with an adnexal mass, 1396 (72%) had benign tumours, 373 (19.2%) had primary invasive tumours, 111 (5.7%) had borderline malignant tumours, and 58 (3%) had metastatic tumours in the ovary. The simple rules yielded a conclusive result in 1501 (77%) masses, for which they resulted in a sensitivity of 92% (95% confidence interval 89% to 94%) and a specificity of 96% (94% to 97%). The corresponding sensitivity and specificity of subjective assessment were 91% (88% to 94%) and 96% (94% to 97%). In the 357 masses for which the simple rules yielded an inconclusive result and with available results of CA-125 measurements, the sensitivities were 89% (83% to 93%) for subjective assessment, 50% (42% to 58%) for the risk of malignancy index, 89% (83% to 93%) for logistic regression model 1, and 82% (75% to 87%) for logistic regression model 2; the corresponding specificities were 78% (72% to 83%), 84% (78% to 88%), 44% (38% to 51%), and 48% (42% to 55%). Use of the simple rules as a triage test and subjective assessment for those masses for which the simple rules yielded an inconclusive result gave a sensitivity of 91% (88% to 93%) and a specificity of 93% (91% to 94%), compared with a sensitivity of 90% (88% to 93%) and a specificity of 93% (91% to 94%) when subjective assessment was used in all masses.
The use of the simple rules has the potential to improve the management of women with adnexal masses. In adnexal masses for which the rules yielded an inconclusive result, subjective assessment of ultrasonic findings by an experienced ultrasound examiner was the most accurate diagnostic test; the risk of malignancy index and the two regression models were not useful.
BMJ (Clinical research ed.). 01/2010; 341:c6839.
-
Caroline Van Holsbeke,
Anneleen Daemen,
Joseph Yazbek,
Tom K Holland,
Tom Bourne,
Tinne Mesens,
Lore Lannoo,
Anne-Sophie Boes,
Annelies Joos,
Arne Van De Vijver,
Nele Roggen,
Bart de Moor,
Eric de Jonge,
Antonia C Testa,
Lil Valentin,
Davor Jurkovic,
Dirk Timmerman
[show abstract]
[hide abstract]
ABSTRACT: To determine how accurately and confidently examiners with different levels of ultrasound experience can classify adnexal masses as benign or malignant and suggest a specific histological diagnosis when evaluating ultrasound images using pattern recognition.
Ultrasound images of selected adnexal masses were evaluated by 3 expert sonologists, 2 senior and 4 junior trainees. They were instructed to classify the masses using pattern recognition as benign or malignant, to state the level of confidence with which this classification was made and to suggest a specific histological diagnosis. Sensitivity, specificity, accuracy and positive and negative likelihood ratios (LR+ and LR-) with regard to malignancy were calculated. The area under the receiver operating characteristic curve (AUC) of pattern recognition was calculated by using six levels of diagnostic confidence.
166 masses were examined, of which 42% were malignant. Sensitivity with regard to malignancy ranged from 80 to 86% for the experts, was 70 and 84% for the 2 senior trainees and ranged from 70 to 86% for the junior trainees. The specificity of the experts ranged from 79 to 91%, was 77 and 89% for the senior trainees and ranged from 59 to 83% for the junior trainees. The experts were uncertain about their diagnosis in 4-13% of the cases, the senior trainees in 15-20% and the junior trainees in 67-100% of the cases. The AUCs ranged from 0.861 to 0.922 for the experts, were 0.842 and 0.855 for the senior trainees, and ranged from 0.726 to 0.795 for the junior trainees. The experts suggested a correct specific histological diagnosis in 69-77% of the cases. All 6 trainees did so significantly less often (22-42% of the cases).
Expert sonologists can accurately classify adnexal masses as benign or malignant and can successfully predict the specific histological diagnosis in many cases. Whilst less experienced operators perform reasonably well when predicting the benign or malignant nature of the mass, they do so with a very low level of diagnostic confidence and are unable to state the likely histology of a mass in most cases.
Gynecologic and Obstetric Investigation 12/2009; 69(3):160-8. · 1.28 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Retroverted uterine incarceration with sacculation of the anterior wall is reported to occur approximately in 1/3000 pregnancies. A literature search identified only one case report of incarceration of an anteflexed gravid uterus and six reported cases of recurrent incarceration and/or sacculation. We present a case of an incarceration of an anteflexed uterus in the first pregnancy, followed by a retroflexed incarceration in the second pregnancy. From this, a review is presented on recurrent uterine incarceration and/or sacculation.
The journal of maternal-fetal & neonatal medicine: the official journal of the European Association of Perinatal Medicine, the Federation of Asia and Oceania Perinatal Societies, the International Society of Perinatal Obstetricians 11/2009; 23(8):776-80. · 1.36 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: To prospectively test the mathematical models for calculation of the risk of malignancy in adnexal masses that were developed on the International Ovarian Tumor Analysis (IOTA) phase 1 data set on a new data set and to compare their performance with that of pattern recognition, our standard method. Methods: Three IOTA centers included 507 new patients who all underwent a transvaginal ultrasound using the standardized IOTA protocol. The outcome measure was the histologic classification of excised tissue. The diagnostic performance of 11 mathematical models that had been developed on the phase 1 data set and of pattern recognition was expressed as area under the receiver operating characteristic curve (AUC) and as sensitivity and specificity when using the cutoffs recommended in the studies where the models had been created. For pattern recognition, an AUC was made based on level of diagnostic confidence.
All IOTA models performed very well and quite similarly, with sensitivity and specificity ranging between 92% and 96% and 74% and 84%, respectively, and AUCs between 0.945 and 0.950. A least squares support vector machine with linear kernel and a logistic regression model had the largest AUCs. For pattern recognition, the AUC was 0.963, sensitivity was 90.2%, and specificity was 92.9%.
This internal validation of mathematical models to estimate the malignancy risk in adnexal tumors shows that the IOTA models had a diagnostic performance similar to that in the original data set. Pattern recognition used by an expert sonologist remains the best method, although the difference in performance between the best mathematical model is not large.
Clinical Cancer Research 02/2009; 15(2):684-91. · 7.74 Impact Factor
-
Caroline Van Holsbeke,
Ben Van Calster,
Stefano Guerriero,
Luca Savelli,
Francesco Leone,
Daniela Fischerova,
Artur Czekierdowski,
Robert Fruscio,
Joan Veldman,
Gregg Van De Putte,
Antonia C Testa,
Tom Bourne,
Lil Valentin,
Dirk Timmerman
[show abstract]
[hide abstract]
ABSTRACT: Aim: To evaluate the performance of subjective evaluation of ultrasound findings (pattern recognition) to discriminate endometriomas from other types of adnexal masses and to compare the demographic and ultrasound characteristics of the true positive cases with those cases that were presumed to be an endometrioma but proved to have a different histology (false positive cases) and the endometriomas missed by pattern recognition (false negative cases). Methods: All patients in the International Ovarian Tumor Analysis (IOTA) studies were included for analysis. In the IOTA studies, patients with an adnexal mass that were preoperatively examined by expert sonologists following the same standardized ultrasound protocol were prospectively included in 21 international centres. Sensitivity and specificity to discriminate endometriomas from other types of adnexal masses using pattern recognition were calculated. Ultrasound and some demographic variables of the masses presumed to be an endometrioma were analysed (true positives and false positives) and compared with the variables of the endometriomas missed by pattern recognition (false negatives) as well as the true negatives. Results: IOTA phase 1, 1b and 2 included 3511 patients of which 2560 were benign (73%) and 951 malignant (27%). The dataset included 713 endometriomas. Sensitivity and specificity for pattern recognition were 81% (577/713) and 97% (2723/2798). The true positives were more often unilocular with ground glass echogenicity than the masses in any other category. Among the 75 false positive cases, 66 were benign but 9 were malignant (5 borderline tumours, 1 rare primary invasive tumour and 3 endometrioid adenocarcinomas). The presumed diagnosis suggested by the sonologist in case of a missed endometrioma was mostly functional cyst or cystadenoma. Conclusion: Expert sonologists can quite accurately discriminate endometriomas from other types of adnexal masses, but in this dataset 1% of the masses that were classified as endometrioma by pattern recognition proved to be malig-nancies.
01/2009; 1:7-17.
-
Neural Computing and Applications. 01/2008; 17:489-500.
-
[show abstract]
[hide abstract]
ABSTRACT: Subjective evaluation of gray-scale and Doppler ultrasound findings (i.e., pattern recognition) by an experienced examiner and preoperative serum levels of CA-125 can both discriminate benign from malignant adnexal (i.e., ovarian, paraovarian, or tubal) masses. We compared the diagnostic performance of these methods in a large multicenter study.
In a prospective multicenter study--the International Ovarian Tumor Analysis--1066 women with a persistent adnexal mass underwent transvaginal gray-scale and color Doppler ultrasound examinations by an experienced examiner within 120 days of surgery. Pattern recognition was used to classify a mass as benign or malignant. Of these women, 809 also had blood collected preoperatively for measurement of serum CA-125. Various levels of CA-125 were used as cutoffs to classify masses. Results from both assays were then compared with histologic findings after surgery.
Pattern recognition correctly classified 93% (95% confidence interval [CI] = 90.9% to 94.6%) of the tumors as benign or malignant. Serum CA-125 correctly classified at best 83% (95% CI = 80.3% to 85.6%) of the masses. Histologic diagnoses that were most often misclassified by CA-125 were fibroma, endometrioma, and abscess (false-positive results) and borderline tumor (false-negative results). Pattern recognition correctly classified 86% (95% CI = 81.1% to 90.4%) of masses of these four histologic types as being benign or malignant, whereas a serum CA-125 at a cutoff of 30 U/mL correctly classified 41% (95% CI = 34.4% to 47.5%) of them. Pattern recognition assigned a correct specific histologic diagnosis to 333 (59%, 95% CI = 54.5% to 62.8%) of the 567 benign lesions.
Pattern recognition was superior to serum CA-125 for discrimination between benign and malignant adnexal masses.
CancerSpectrum Knowledge Environment 12/2007; 99(22):1706-14. · 14.07 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: To test the value of serum CA-125 measurements alone or as part of a multimodal strategy to distinguish between malignant and benign ovarian tumors before surgery based on a large prospective multicenter study (International Ovarian Tumor Analysis).
Patients with at least one persistent ovarian mass preoperatively underwent transvaginal ultrasonography using gray scale imaging to assess tumor morphology and color Doppler imaging to obtain indices of blood flow.
Data from 809 patients recruited from nine centers were included in the analysis; 567 patients (70%) had benign tumors and 242 (30%) had malignant tumors-of these 152 were primary invasive (62.8%), 52 were borderline malignant (21.5%), and 38 were metastatic (15.7%). A logistic regression model including CA-125 (M2) resulted in an area under the receiver operating characteristic curve (AUC) of 0.934 and did not outperform a published (M1) without serum CA-125 information (AUC, 0.936). Specifically designed new models including CA-125 for premenopausal women (M3) and for postmenopausal women (M4) did not perform significantly better than the model without CA-125 (M1; AUC, 0.891 v AUC, 0.911 and AUC, 0.975 v AUC, 0.949, respectively). In postmenopausal patients, serum CA-125 alone (AUC, 0.920) and the risk of malignancy index (AUC, 0.924) performed very well. Results were very similar when the models were prospectively tested on a group of 345 new patients with adnexal masses of whom 126 had malignant tumors (37%).
Adding information on CA-125 to clinical information and ultrasound information does not improve discrimination of mathematical models between benign and malignant adnexal masses.
Journal of Clinical Oncology 10/2007; 25(27):4194-200. · 18.37 Impact Factor
-
Caroline Van Holsbeke,
Ben Van Calster,
Lil Valentin,
Antonia C Testa,
Enrico Ferrazzi,
Ioannis Dimou,
Chuan Lu,
Philippe Moerman,
Sabine Van Huffel,
Ignace Vergote,
Dirk Timmerman
[show abstract]
[hide abstract]
ABSTRACT: Several scoring systems have been developed to distinguish between benign and malignant adnexal tumors. However, few of them have been externally validated in new populations. Our aim was to compare their performance on a prospectively collected large multicenter data set.
In phase I of the International Ovarian Tumor Analysis multicenter study, patients with a persistent adnexal mass were examined with transvaginal ultrasound and color Doppler imaging. More than 50 end point variables were prospectively recorded for analysis. The outcome measure was the histologic classification of excised tissue as malignant or benign. We used the International Ovarian Tumor Analysis data to test the accuracy of previously published scoring systems. Receiver operating characteristic curves were constructed to compare the performance of the models.
Data from 1,066 patients were included; 800 patients (75%) had benign tumors and 266 patients (25%) had malignant tumors. The morphologic scoring system used by Lerner gave an area under the receiver operating characteristic curve (AUC) of 0.68, whereas the multimodal risk of malignancy index used by Jacobs gave an AUC of 0.88. The corresponding values for logistic regression and artificial neural network models varied between 0.76 and 0.91 and between 0.87 and 0.90, respectively. Advanced kernel-based classifiers gave an AUC of up to 0.92.
The performance of the risk of malignancy index was similar to that of most logistic regression and artificial neural network models. The best result was obtained with a relevance vector machine with radial basis function kernel. Because the models were tested on a large multicenter data set, results are likely to be generally applicable.
Clinical Cancer Research 09/2007; 13(15 Pt 1):4440-7. · 7.74 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Ovarian masses are common and a good pre-surgical assessment of their nature is important for adequate treatment. Bayesian Multi-Layer Perceptrons (MLPs) using the evidence procedure were used to predict whether tumors are malignant or not. Automatic Relevance Determination (ARD) is used to select the most relevant of the 40+ available variables. Cross-validation is used to select an optimal combination of input set and number of hidden neurons. The data set consists of 1066 tumors collected at nine centers across Europe. Results indicate good performance of the models with AUC values of 0.93-0.94 on independent data. A comparison with a Bayesian perceptron model shows that the present problem is to a large extent linearly separable. The analyses further show that the number of hidden neurons specified in the ARD analyses for input selection may influence model performance.
Conference proceedings: ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Conference 02/2006; 1:5342-5.
