Risk of complications in patients with conservatively managed ovarian tumours (IOTA5): a 2-year interim analysis of a multicentre, prospective, cohort study

Deep plasma proteomics identifies and validates an eight-protein biomarker panel that separate benign from malignant tumors in ovarian cancer

BACKGROUND

Ovarian cancer has the highest mortality of all gynecological cancers and surgery is commonly used as final diagnostic. Available literature indicates that women with benign tumors could often be conservatively managed, but accurate molecular tests are needed for triaging when gold-standard imaging techniques are inconclusive or lacking.

METHODS

Here, we analyzed 5416 plasma proteins in two independent cohorts (N1 = 171, N2 = 233) with women surgically diagnosed with benign or malignant tumors. Using one cohort as discovery, we compared protein levels of benign tumors with early stage (I-II), late stage (III-IV) or any stage (I-IV) ovarian cancer and trained risk-score reporting multivariate models including a fixed cut-off for malignancy. Associations and model performance was then evaluated in the replication cohort.

RESULTS

We identify 327 biomarker associations, corresponding to 191 unique proteins, and replicate 326 (99.7%). By comparing the 191 proteins with their corresponding tumor gene expression we find that only 11% (21/191) have significant correlation. Through analyzes of protein-protein correlation networks, we find that 62 of the 191 proteins have high correlation with at least one other protein, suggesting that many of the associations are secondary effects. In the replication cohort, our model has areas under the curve (AUC = 0.96) corresponding to 97% sensitivity at 68% specificity. For early-stage tumors, we estimate the sensitivity to 91% at a specificity of 68% as compared to 85% and 54% for CA-125 alone.

CONCLUSIONS

Our results indicates that up to one third of benign cases can be identified by molecular measures thereby reducing the need for diagnostic surgery.

Accurate prediction of benign and malignant adnexal tumors in surgical resection and conservative treatment: construction and external validation of a diagnostic model based on CEUS, HE4, and O-RADS US v2022 evaluation

PURPOSE

To establish a diagnostic model combining contrast-enhanced ultrasound (CEUS), human epididymis protein 4 (HE4), and Ovarian-Adnexal Reporting and Data Systems (O-RADS) US v2022, verify its diagnostic efficacy, and compare it with subjective evaluation.

METHODS

From January 2018 to August 2021 (the test group) and from September 2021 to September 2022 (the validation group), the data of patients classified as O-RADS US v2022 categories 2 to 5 who underwent adnexal ultrasound examinations were prospectively and continuously collected. In the test group, univariate and multivariate analyses were used to explore the relationship between age, body mass index (BMI), maximum diameter of the lesion, menopausal status, HE4, cancer antigen 125 (CA125), and the characteristics of CEUS and malignant lesions. Selecting independent influencing factors to construct diagnostic model, which was validated in the external validation group and compared with subjective evaluation.

RESULTS

The test group included 563 patients (mean age, 48.7 ± 13.2), and the validation group included 246 patients (mean age, 47.6 ± 12.9). Univariate and multivariate analyses showed that enhancement time, enhancement intensity, dynamic changes, and HE4 were independent influencing factors for predicting adnexal malignant tumors. In the validation group, the sensitivities and specificities of O-RADS US v2022, O-RADS US v2022 + CEUS, O-RADS US v2022 + CEUS + HE4, and subjective assessment were 88.89% and 70.69%, 94.44% and 79.31%, 91.67% and 92.53%, and 93.09% and 89.66% respectively. In addition, the combined diagnostic performance of O-RADS US v2022, CEUS and HE4 (AUC = 0.980) was higher than that of O-RADS US v2022 alone (AUC = 0.876, P < 0.001) and the combination of O-RADS US v2022 + CEUS (AUC = 0.908, P < 0.001), and was comparable to the subjective evaluation (AUC = 0.963, P = 0.192).

CONCLUSIONS

The combined diagnostic model of O-RADS US v2022, CEUS and HE4 can improve the specificity of adnexal ultrasound diagnosis without sacrificing sensitivity, and it has high reliability.

External validation of the IOTA two-step strategy in the preoperative characterization of ovarian masses

OBJECTIVES

Preoperative sonographic evaluation of ovarian masses is crucial for improving outcomes. The Risk of Malignancy Index (RMI) has been a standard for malignancy triage, while the International Ovarian Tumor Analysis Group (IOTA) has proposed a two-step strategy to estimate the risk of malignancy and suggest management steps by translating risks to Ovarian Adnexal Reporting Data System (O-RADS) categories. This study compares the accuracy of RMI and the IOTA two-step strategy in predicting malignancy.

METHODS

We included patients with preoperative ultrasound and pathological reports. RMI and O-RADS scores based on the IOTA two-step strategy were assessed. Performance was evaluated using receiver operating characteristic (ROC) curves and calibration plots.

RESULTS

A total of 453 cases were included. Of these, 90 (19.9 %) were malignant, 21 (4.6 %) were borderline tumors (BOT), and 342 (75.5 %) were benign. The area under the ROC curve (AUC) for the IOTA two-step strategy was 0.958 (95 % CI, 0.938-0.978), compared to 0.904 (0.865-0.943) for RMI with a > 200 cut-off. The IOTA two-step strategy had a sensitivity of 96.4 %, specificity of 79.7 %, positive predictive value (PPV) 60.2 %, and negative predictive value (NPV) 98.6 %, while RMI showed sensitivity of 70.4 %, specificity 93.4 %, PPV 79.2 %, and NPV 89.8 %. For predicting BOTs, the IOTA two-step AUC was 0.902, compared to 0.719 for RMI.

CONCLUSION

The IOTA two-step strategy outperforms RMI in the preoperative assessment of adnexal masses, particularly in detecting BOTs. It should be implemented in routine clinical practice.

Multimodal ultrasound-based radiomics and deep learning for differential diagnosis of O-RADS 4-5 adnexal masses

BACKGROUND

Accurate differentiation between benign and malignant adnexal masses is crucial for patients to avoid unnecessary surgical interventions. Ultrasound (US) is the most widely utilized diagnostic and screening tool for gynecological diseases, with contrast-enhanced US (CEUS) offering enhanced diagnostic precision by clearly delineating blood flow within lesions. According to the Ovarian and Adnexal Reporting and Data System (O-RADS), masses classified as categories 4 and 5 carry the highest risk of malignancy. However, the diagnostic accuracy of US remains heavily reliant on the expertise and subjective interpretation of radiologists. Radiomics has demonstrated significant value in tumor differential diagnosis by extracting microscopic information imperceptible to the human eye. Despite this, no studies to date have explored the application of CEUS-based radiomics for differentiating adnexal masses. This study aims to develop and validate a multimodal US-based nomogram that integrates clinical variables, radiomics, and deep learning (DL) features to effectively distinguish adnexal masses classified as O-RADS 4-5.

METHODS

From November 2020 to March 2024, we enrolled 340 patients who underwent two-dimensional US (2DUS) and CEUS and had masses categorized as O-RADS 4-5. These patients were randomly divided into a training cohort and a test cohort in a 7:3 ratio. Adnexal masses were manually segmented from 2DUS and CEUS images. Using machine learning (ML) and DL techniques, five models were developed and validated to differentiate adnexal masses. The diagnostic performance of these models was evaluated using the area under the receiver operating characteristic (ROC) curve (AUC), accuracy, sensitivity, specificity, precision, and F1-score. Additionally, a nomogram was constructed to visualize outcome measures.

RESULTS

The CEUS-based radiomics model outperformed the 2DUS model (AUC: 0.826 vs. 0.737). Similarly, the CEUS-based DL model surpassed the 2DUS model (AUC: 0.823 vs. 0.793). The ensemble model combining clinical variables, radiomics, and DL features achieved the highest AUC (0.929).

CONCLUSIONS

Our study confirms the effectiveness of CEUS-based radiomics for distinguishing adnexal masses with high accuracy and specificity using a multimodal US-based radiomics DL nomogram. This approach holds significant promise for improving the diagnostic precision of adnexal masses classified as O-RADS 4-5.

Endometriosis: assessment on O-RADS and risk of malignant transformation

Endometriosis is a common disease, affecting approximately 10% of women of reproductive age. Several intersecting guidelines and consensus statements provide information on imaging diagnosis and surveillance strategies for endometriomas. SRU consensus panel recommendations provide information on initial detection of endometriosis on routine pelvic imaging. Revised American Society of Reproductive Medicine (rASRM) classification, the #ENZIAN classification, and the deep pelvic endometriosis index (dPEI) aim to assess the overall extent of disease and assist in presurgical planning. The Ovarian-Adnexal Reporting and Data System (O-RADS) aims to risk stratify lesions evaluated with US or MR based on their imaging morphology, from typical benign lesions to atypical presentations and malignant transformation. Emerging data shows increased risk of ovarian cancer in patients with endometriosis, especially following menopause and in those patients with long standing endometriosis. (Chen et al. in Front Oncol. 14:1329133, 2024;Streuli et al. in Climacteric. 20:138-143, 2017;Secosan et al. in Diagnostics (Basel). 10:134, 2020;Inceboz in Womens Health (Lond Engl). 11:711-715, 2015;Cassani et al. in Maturitas. 190, 2024;Gemmell et al. in Hum Reprod Update. 23:481-500, 2017;Giannella et al. in Cancers (Basel). 13:4026, 2021;) Current O-RADS guidelines mandate follow-up of endometriomas up to 2 years with further follow-up based on clinical factors. No consensus guidelines exist for imaging surveillance of patients with deep endometriosis from a malignancy standpoint. This review explores the imaging appearance of endometriomas, imaging features of malignant transformation, surveillance strategies and gaps in current literature, and attempts to better understand the risk of malignancy and to encourage further research for long-term imaging surveillance of endometriosis patients.

Radiomics analysis of ultrasound images to discriminate between benign and malignant adnexal masses with solid morphology on ultrasound

OBJECTIVE

The primary aim was to identify radiomics ultrasound features that can distinguish between benign and malignant adnexal masses with solid ultrasound morphology, and between primary malignant (including borderline and primary invasive) and metastatic solid ovarian masses, and to develop ultrasound-based machine learning models that include radiomics features to discriminate between benign and malignant solid adnexal masses. The secondary aim was to compare the discrimination performance of our newly developed radiomics models with that of the Assessment of Different NEoplasias in the adneXa (ADNEX) model and that of subjective assessment by an experienced ultrasound examiner.

METHODS

This was a retrospective, observational single-center study conducted at Fondazione Policlinico Universitario A. Gemelli IRCC, in Rome, Italy. Included were patients with a histological diagnosis of an adnexal tumor with solid morphology according to International Ovarian Tumor Analysis (IOTA) terminology at preoperative ultrasound examination performed in 2014-2020, who were managed with surgery. The patient cohort was split randomly into training and validation sets at a ratio of 70:30 and with the same proportion of benign and malignant tumors in the two subsets, with malignant tumors including borderline, primary invasive and metastatic tumors. We extracted 68 radiomics features, belonging to two different families: intensity-based statistical features and textural features. Models to predict malignancy were built based on a random forest classifier, fine-tuned using 5-fold cross-validation over the training set, and tested on the held-out validation set. The variables used in model-building were patient age and radiomics features that were statistically significantly different between benign and malignant adnexal masses and assessed as not redundant based on the Pearson correlation coefficient. We evaluated the discriminative ability of the models and compared it to that of the ADNEX model and that of subjective assessment by an experienced ultrasound examiner using the area under the receiver-operating-characteristics curve (AUC) and classification performance by calculating sensitivity and specificity.

RESULTS

In total, 326 patients were included and 775 preoperative ultrasound images were analyzed. Of the 68 radiomics features extracted, 52 differed statistically significantly between benign and malignant tumors in the training set, and 18 uncorrelated features were selected for inclusion in model-building. The same 52 radiomics features differed significantly between benign, primary malignant and metastatic tumors. However, the values of the features manifested overlapped between primary malignant and metastatic tumors and did not differ significantly between them. In the validation set, 25/98 (25.5%) tumors were benign and 73/98 (74.5%) were malignant (6 borderline, 57 primary invasive, 10 metastatic). In the validation set, a model including only radiomics features had an AUC of 0.80, sensitivity of 0.78 and specificity of 0.76 at an optimal cut-off for risk of malignancy of 68%, based on Youden's index. The corresponding results for a model including age and radiomics features were AUC of 0.79, sensitivity of 0.86 and specificity of 0.56 (cut-off 60%, based on Youden's index), while those of the ADNEX model were AUC of 0.88, sensitivity of 0.99 and specificity of 0.64 (at a 20% risk-of-malignancy cut-off). Subjective assessment had a sensitivity of 0.99 and specificity of 0.72.

CONCLUSIONS

Our radiomics model had moderate discriminative ability on internal validation and the addition of age to this model did not improve its performance. Even though our radiomics models had discriminative ability inferior to that of the ADNEX model, our results are sufficiently promising to justify continued development of radiomics analysis of ultrasound images of adnexal masses. © 2024 The Author(s). Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.

Performance of grayscale combined with contrast-enhanced ultrasound in differentiating benign and malignant pediatric ovarian masses

OBJECTIVES

To evaluate grayscale US combined with contrast-enhanced ultrasound (CEUS) in the preoperative differentiation between benign and malignant ovarian masses in a pediatric population.

MATERIALS AND METHODS

This retrospective study enrolled patients who underwent grayscale US and CEUS before surgery because of ovarian masses between July 2018 and September 2023, with available histopathologic or follow-up results. Two senior radiologists summarized the grayscale US and CEUS characteristics of all ovarian masses, including percentage of solidity, ascites, vascularity, and enhanced vessel morphology. These characteristics were then independently reviewed by radiologists with different experience to assess interobserver agreement. Diagnostic performance was evaluated using the area under the receiver operating characteristic curve (AUC), while interobserver agreement was evaluated by intraclass correlation coefficient (ICC).

RESULTS

A total of 26 children (median age: 10.1 [7.5, 11.7] years; age range: 0-14 years; benign: 15 patients) were included. The main characteristics of malignant ovarian tumors were abundant blood flow and twisted blood vessels within the mass, enhanced portion of the mass over 50 percent (all p < 0.001). The grayscale US combined with CEUS showed better diagnostic performance than the grayscale US alone (AUC = 0.99 [95% CI: 0.95, 1.00] vs AUC = 0.70 [95% CI: 0.50, 0.90] p < 0.001). A statistically significant AUC before and after CEUS was also shown between two junior radiologists (0.75 vs 0.92 and 0.69 vs 0.86, respectively, both p < 0.05). ICC of CEUS was better than that of grayscale US among radiologists.

CONCLUSION

The combination of grayscale US and CEUS might improve the diagnostic accuracy in differentiating benign and malignant pediatric ovarian masses.

CLINICAL RELEVANCE STATEMENT

Grayscale ultrasound combined with contrast-enhanced ultrasound can improve the diagnostic performance in the preoperative differentiation of benign and malignant ovarian lesions in a pediatric population.

KEY POINTS

Correctly distinguishing benign from malignant ovarian masses in pediatric patients is critical for determining treatments. Grayscale combined with contrast-enhanced ultrasound (CEUS) differentiated benign and malignant pediatric ovarian masses better than grayscale US alone. Junior radiologists' diagnostic performances could be and were significantly improved with the application of CEUS.

Diagnostic value of the International Ovarian Tumor Analysis simple rules combined with contrast-enhanced ultrasound for adnexal masses

OBJECTIVE

This study aimed to evaluate the diagnostic accuracy of the International Ovarian Tumor Analysis (IOTA) simple rules combined with contrast-enhanced ultrasound (CEUS) in differentiating benign from malignant adnexal masses.

METHODS

This retrospective study included 179 patients with adnexal masses who underwent pre-operative ultrasound. The IOTA simple rules were applied first, followed by CEUS for inconclusive or suspicious cases. Diagnostic accuracy, sensitivity, specificity, positive predictive value, and negative predictive value were calculated for the IOTA simple rules alone and combined with CEUS.

RESULTS

Among the 179 adnexal masses, 113 (63.1%) were benign and 66 (36.9%) were malignant or borderline. The IOTA simple rules alone achieved 86.8% sensitivity, 95.6% specificity, and 92.3% accuracy. When combined with CEUS, sensitivity increased to 92.7%, specificity to 98.3%, and accuracy to 96.2%. Sub-group analysis showed that the combined approach was particularly beneficial in women who were pre-menopausal, with sensitivity rising from 71.4% to 85.7%.

CONCLUSION

Combining the IOTA simple rules with CEUS significantly improves diagnostic accuracy in distinguishing benign from malignant adnexal masses, especially in inconclusive cases. This approach may enhance management across age groups.

A Study on Automatic O-RADS Classification of Sonograms of Ovarian Adnexal Lesions Based on Deep Convolutional Neural Networks

OBJECTIVE

This study explored a new method for automatic O-RADS classification of sonograms based on a deep convolutional neural network (DCNN).

METHODS

A development dataset (DD) of 2,455 2D grayscale sonograms of 870 ovarian adnexal lesions and an intertemporal validation dataset (IVD) of 426 sonograms of 280 lesions were collected and classified according to O-RADS v2022 (categories 2-5) by three senior sonographers. Classification results verified by a two-tailed z-test to be consistent with the O-RADS v2022 malignancy rate indicated the diagnostic performance was comparable to that of a previous study and were used for training; otherwise, the classification was repeated by two different sonographers. The DD was used to develop three DCNN models (ResNet34, DenseNet121, and ConvNeXt-Tiny) that employed transfer learning techniques. Model performance was assessed for accuracy, precision, and F1 score, among others. The optimal model was selected and validated over time using the IVD and to analyze whether the efficiency of O-RADS classification was improved with the assistance of this model for three sonographers with different years of experience.

RESULTS

The proportion of malignant tumors in the DD and IVD in each O-RADS-defined risk category was verified using a two-tailed z-test. Malignant lesions (O-RADS categories 4 and 5) were diagnosed in the DD and IVD with sensitivities of 0.949 and 0.962 and specificities of 0.892 and 0.842, respectively. ResNet34, DenseNet121, and ConvNeXt-Tiny had overall accuracies of 0.737, 0.752, and 0.878, respectively, for sonogram prediction in the DD. The ConvNeXt-Tiny model's accuracy for sonogram prediction in the IVD was 0.859, with no significant difference between test sets. The modeling aid significantly reduced O-RADS classification time for three sonographers (Cohen's d = 5.75).

CONCLUSION

ConvNeXt-Tiny showed robust and stable performance in classifying O-RADS 2-5, improving sonologists' classification efficacy.

Diagnostic performance of a modified O-RADS classification system for adnexal lesions incorporating clinical features

PURPOSE

To compare the diagnostic efficacy of the Ovarian-Adnexal Reporting and Data System (O-RADS) MRI score with that of the modified O-RADS score on the basis of a simplified contrast-enhanced (CE) MRI protocol in characterizing adnexal masses with solid tissue. The added value of clinical features was evaluated to improve the ability of the scoring system to classify adnexal masses.

METHODS

A total of 124 patients with 124 adnexal lesions containing solid tissue were included in this two-center retrospective study. Among them, there were 40 benign lesions (40/124, 32.3%) and 84 were malignant lesions (84/124, 67.7%). Three radiologists independently reviewed the images and assigned the O-RADS MRI score and the modified O-RADS score for each adnexal mass. Histopathology was used as the reference standard. The diagnostic efficacy of the two scoring methods was compared. Univariate and multivariate logistic regression were performed to evaluate the value of significant features in the prediction of malignant tumors.

RESULTS

The O-RADS MRI score and modified O-RADS score showed sensitivity at 100.0% (95% CI, 95.7-100.0%) and 71.4% (95% CI, 60.5-80.8%), specificity at 12.5% (95% CI, 4.2-26.8%) and 75.0% (95% CI, 58.8-87.3%), respectively. The area under the curve of the modified O-RADS score was higher than the O-RADS score (0.732 [95% CI, 0.645-0.808] vs 0.575 [95% CI, 0.483-0.663]; p < 0.001). Multivariate analysis showed that the modified O-RADS score 4b or 5 combined with patient age > 38.5 years, nullipara, maximum diameter > 40.5 mm and HE4 > 78.9 pmol/L significantly improved the diagnostic efficacy up to 0.954 (95% CI, 0.901-0.984) (p < 0.001).

CONCLUSION

A modified O-RADS score combined with certain clinical features can significantly improve the diagnostic efficacy in predicting malignant tumors.

Inter-reader reliability of Ovarian-Adnexal Reporting and Data System US: a systematic review and meta-analysis

PURPOSE

Ovarian-Adnexal Reporting and Data System (O-RADS) US provides a standardized lexicon for ovarian and adnexal lesions, facilitating risk stratification based on morphological features for malignancy assessment, which is essential for proper management. However, systematic determination of inter-reader reliability in O-RADS US categorization remains unexplored. This study aimed to systematically determine the inter-reader reliability of O-RADS US categorization and identify the factors that affect it.

METHODS

Original articles reporting the inter-reader reliability of O-RADS US in lesion categorization were identified in the MEDLINE, EMBASE, and Web of Science databases from January 2018 to December 2023. DerSimonian-Laird random-effects models were used to determine the meta-analytic pooled inter-reader reliability of the O-RADS US categorization. Subgroup meta-regression analysis was performed to identify the factors causing study heterogeneity.

RESULTS

Fourteen original articles with 5139 ovarian and adnexal lesions were included. The inter-reader reliability of O-RADS US in lesion categorization ranged from 0.71 to 0.99, with a meta-analytic pooled estimate of 0.83 (95% CI, 0.78-0.88), indicating almost perfect reliability. Substantial study heterogeneity was observed in the inter-reader reliability of the O-RADS US categorization (I2 = 96.9). In subgroup meta-regression analysis, reader experience was the only factor associated with study heterogeneity. Pooled inter-reader reliability of the O-RADS US categorization was higher in studies with all experienced readers (0.86; 95% CI, 0.81-0.91) compared to those with multiple readers including trainees (0.74; 95% CI, 0.70-0.78; P = 0.009). The inter-reader reliability of US descriptors ranged from 0.39 to 0.97, with ascites and peritoneal nodules showing almost perfect reliability (0.79- 0.97).

CONCLUSION

The O-RADS US risk stratification system demonstrated almost perfect inter-reader reliability in lesion categorization. Our results highlight the importance of targeted training and descriptor simplification to improve inter-reader reliability and clinical adoption.

The diagnostic performance of International Ovarian Tumor Analysis: Simple Rules for diagnosing ovarian tumors-a systematic review and meta-analysis

INTRODUCTION

Adnexal masses are a common health issue in gynecology; the challenge lies in the differential diagnosis of these masses. The International Ovarian Tumor Analysis Simple Rules (IOTA-SR) offers the first scoring system to aid in diagnosis. It is based on a set of five ultrasound imaging features indicative of a malignant ovarian tumor and five features indicative of a benign tumor. This review aims to assess the diagnostic performance of IOTA-SR for classifying ovarian tumors as benign or malignant.

METHODS

A systematic review was conducted on MEDLINE, Embase, Google Scholar, Scopus, and Web of Science. The terminologies "IOTA-SR", "adnexal, mass", and "ovarian tumors scoring" were employed. Twenty-seven research articles conducted from 2008 to 2022 were included in the meta-analysis; the publication outcome indicates that performance quality tests were extracted directly or indirectly, including true positive (TP), false positive (FP), true negative (TN), and false negative (FN). The Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) was used to evaluate the study quality and estimate the risk of bias. After estimating the pooled effect of the sensitivity, specificity, and diagnostic odds ratio (DOR), the summary receiver operating characteristic (SROC) curve was estimated using the bivariate random effects model. Utilizing Cochran's Q statistics and Higgins's inconsistency test through the I2 index for pooled analysis, the heterogeneity of studies was quantitatively evaluated. The funnel plot and Egger's test were utilized to visually and quantitatively evaluate potential publication bias.

RESULTS

Among 27 studies, including 7,841 adnexal masses, the results of this meta-analysis showed excellent diagnostic performance with a pooled sensitivity of 92% [95% confidence interval (CI), 0.89-0.94] and a pooled specificity of 92% (95% CI, 0.89-0.94). The IOTA-SR was applicable in 85.7% of adnexal masses.

CONCLUSION

The IOTA-SR is highly effective in the presurgical differentiation of malignant versus benign adnexal masses when applied by an expert ultrasonography operator.

Automatic segmentation model and machine learning model grounded in ultrasound radiomics for distinguishing between low malignant risk and intermediate-high malignant risk of adnexal masses

OBJECTIVE

To develop an automatic segmentation model to delineate the adnexal masses and construct a machine learning model to differentiate between low malignant risk and intermediate-high malignant risk of adnexal masses based on ovarian-adnexal reporting and data system (O-RADS).

METHODS

A total of 663 ultrasound images of adnexal mass were collected and divided into two sets according to experienced radiologists: a low malignant risk set (n = 446) and an intermediate-high malignant risk set (n = 217). Deep learning segmentation models were trained and selected to automatically segment adnexal masses. Radiomics features were extracted utilizing a feature analysis system in Pyradiomics. Feature selection was conducted using the Spearman correlation analysis, Mann-Whitney U-test, and least absolute shrinkage and selection operator (LASSO) regression. A nomogram integrating radiomic and clinical features using a machine learning model was established and evaluated. The SHapley Additive exPlanations were used for model interpretability and visualization.

RESULTS

The FCN ResNet101 demonstrated the highest segmentation performance for adnexal masses (Dice similarity coefficient: 89.1%). Support vector machine achieved the best AUC (0.961, 95% CI: 0.925-0.996). The nomogram using the LightGBM algorithm reached the best AUC (0.966, 95% CI: 0.927-1.000). The diagnostic performance of the nomogram was comparable to that of experienced radiologists (p > 0.05) and outperformed that of less-experienced radiologists (p < 0.05). The model significantly improved the diagnostic accuracy of less-experienced radiologists.

CONCLUSIONS

The segmentation model serves as a valuable tool for the automated delineation of adnexal lesions. The machine learning model exhibited commendable classification capability and outperformed the diagnostic performance of less-experienced radiologists.

CRITICAL RELEVANCE STATEMENT

The ultrasound radiomics-based machine learning model holds the potential to elevate the professional ability of less-experienced radiologists and can be used to assist in the clinical screening of ovarian cancer.

KEY POINTS

We developed an image segmentation model to automatically delineate adnexal masses. We developed a model to classify adnexal masses based on O-RADS. The machine learning model has achieved commendable classification performance. The machine learning model possesses the capability to enhance the proficiency of less-experienced radiologists. We used SHapley Additive exPlanations to interpret and visualize the model.

International multicenter validation of AI-driven ultrasound detection of ovarian cancer

Ovarian lesions are common and often incidentally detected. A critical shortage of expert ultrasound examiners has raised concerns of unnecessary interventions and delayed cancer diagnoses. Deep learning has shown promising results in the detection of ovarian cancer in ultrasound images; however, external validation is lacking. In this international multicenter retrospective study, we developed and validated transformer-based neural network models using a comprehensive dataset of 17,119 ultrasound images from 3,652 patients across 20 centers in eight countries. Using a leave-one-center-out cross-validation scheme, for each center in turn, we trained a model using data from the remaining centers. The models demonstrated robust performance across centers, ultrasound systems, histological diagnoses and patient age groups, significantly outperforming both expert and non-expert examiners on all evaluated metrics, namely F1 score, sensitivity, specificity, accuracy, Cohen's kappa, Matthew's correlation coefficient, diagnostic odds ratio and Youden's J statistic. Furthermore, in a retrospective triage simulation, artificial intelligence (AI)-driven diagnostic support reduced referrals to experts by 63% while significantly surpassing the diagnostic performance of the current practice. These results show that transformer-based models exhibit strong generalization and above human expert-level diagnostic accuracy, with the potential to alleviate the shortage of expert ultrasound examiners and improve patient outcomes.

Radiomics and radiogenomics: extracting more information from medical images for the diagnosis and prognostic prediction of ovarian cancer

Ovarian cancer (OC) remains one of the most lethal gynecological malignancies globally. Despite the implementation of various medical imaging approaches for OC screening, achieving accurate differential diagnosis of ovarian tumors continues to pose significant challenges due to variability in image performance, resulting in a lack of objectivity that relies heavily on the expertise of medical professionals. This challenge can be addressed through the emergence and advancement of radiomics, which enables high-throughput extraction of valuable information from conventional medical images. Furthermore, radiomics can integrate with genomics, a novel approach termed radiogenomics, which allows for a more comprehensive, precise, and personalized assessment of tumor biological features. In this review, we present an extensive overview of the application of radiomics and radiogenomics in diagnosing and predicting ovarian tumors. The findings indicate that artificial intelligence methods based on imaging can accurately differentiate between benign and malignant ovarian tumors, as well as classify their subtypes. Moreover, these methods are effective in forecasting survival rates, treatment outcomes, metastasis risk, and recurrence for patients with OC. It is anticipated that these advancements will function as decision-support tools for managing OC while contributing to the advancement of precision medicine.

Evaluating use of two-step International Ovarian Tumor Analysis strategy to classify adnexal masses identified in pregnancy: pilot study

OBJECTIVES

The primary aim was to validate the International Ovarian Tumor Analysis (IOTA) benign simple descriptors (BDs) followed by the Assessment of Different NEoplasias in the adneXa (ADNEX) model, if BDs cannot be applied, in a two-step strategy to classify adnexal masses identified during pregnancy. The secondary aim was to describe the natural history of adnexal masses during pregnancy.

METHODS

This was a retrospective analysis of prospectively collected data from women with an adnexal mass identified on ultrasonography during pregnancy between 2017 and 2022 at Queen Charlotte's and Chelsea Hospital, London, UK. Clinical and ultrasound data were extracted from medical records and ultrasound software. Adnexal masses were classified and managed according to expert subjective assessment (SA). Borderline ovarian tumors (BOTs) were classified as malignant. BDs were applied retrospectively to classify adnexal masses, and if BDs were not applicable, the ADNEX model (using a risk- of-malignancy threshold ≥ 10%) was used, in a two-step strategy. The reference standard was histology (where available) or expert SA at the postnatal ultrasound scan.

RESULTS

A total of 291 women with a median age of 33 (interquartile range (IQR), 29-36) years presented with an adnexal mass during pregnancy, at a median gestational age of 12 (IQR, 8-17) weeks. Of those, 267 (91.8%) were followed up to the postnatal period. Based on the reference standard, 4.1% (11/267) of adnexal masses were classified as malignant (all BOTs) and 95.9% (256/267) as benign. BDs were applicable in 68.9% (184/267) of adnexal masses; of these, only one (0.5%) BOT was misclassified as benign. The ADNEX model was used to classify the 83 residual masses and misclassified 3/10 (30.0%) BOTs as benign and 25/73 (34.2%) benign masses as malignant, of which 13/25 (52.0%) were classified as decidualized endometrioma on expert SA. The two-step strategy had a specificity of 90.2%, sensitivity of 63.6%, negative predictive value of 98.3% and positive predictive value of 21.9%. A total of 56 (21.0%) women underwent surgical intervention: four (1.5%) as an emergency during pregnancy, four (1.5%) electively during Cesarean section and 48 (18.0%) postnatally. During follow-up, 64 (24.0%) adnexal masses resolved spontaneously. Cyst-related complications occurred in four (1.5%) women during pregnancy (ovarian torsion, n = 2; cyst rupture, n = 2) and six (2.2%) women in the postnatal period (all ovarian torsion). Overall, 196/267 (73.4%) women had a persistent adnexal mass at postnatal ultrasound. Presumed decidualization occurred in 31.1% (19/61) of endometriomas and had resolved in 89.5% (17/19) by the first postnatal ultrasound scan.

CONCLUSIONS

BDs apply to most adnexal masses during pregnancy. However, the small number of malignant tumors in this cohort (4.1%) restricted the evaluation of the ADNEX model, so expert SA should be used to classify adnexal masses during pregnancy when BDs do not apply. A larger multicenter prospective study is required to evaluate the use of the ADNEX model to classify adnexal masses during pregnancy. Our data suggest that most adnexal masses can be managed expectantly during pregnancy, given the high rate of spontaneous resolution and low risk of complications. © 2024 The Author(s). Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.

Development and validation of a deep learning pipeline to diagnose ovarian masses using ultrasound screening: a retrospective multicenter study

Background

Ovarian cancer has the highest mortality rate among gynaecological malignancies and is initially screened using ultrasound. Owing to the high complexity of ultrasound images of ovarian masses and the anatomical characteristics of the deep pelvic cavity, subjective assessment requires extensive experience and skill. Therefore, detecting the ovaries and ovarian masses and diagnose ovarian cancer are challenging. In the present study, we aimed to develop an automated deep learning framework, the Ovarian Multi-Task Attention Network (OvaMTA), for ovary and ovarian mass detection, segmentation, and classification, as well as further diagnosis of ovarian masses based on ultrasound screening.

Methods

Between June 2020 and May 2022, the OvaMTA model was trained, validated and tested on a training and validation cohort including 6938 images and an internal testing cohort including 1584 images which were recruited from 21 hospitals involving women who underwent ultrasound examinations for ovarian masses. Subsequently, we recruited two external test cohorts from another two hospitals. We obtained 1896 images between February 2024 and April 2024 as image-based external test dataset, and further obtained 159 videos for the video-based external test dataset between April 2024 and May 2024. We developed an artificial intelligence (AI) system (termed OvaMTA) to diagnose ovarian masses using ultrasound screening. It includes two models: an entire image-based segmentation model, OvaMTA-Seg, for ovary detection and a diagnosis model, OvaMTA-Diagnosis, for predicting the pathological type of ovarian mass using image patches cropped by OvaMTA-Seg. The performance of the system was evaluated in one internal and two external validation cohorts, and compared with doctors' assessments in real-world testing. We recruited eight physicians to assess the real-world data. The value of the system in assisting doctors with diagnosis was also evaluated.

Findings

In terms of segmentation, OvaMTA-Seg achieved an average Dice score of 0.887 on the internal test set and 0.819 on the image-based external test set. OvaMTA-Seg also performed well in ovarian mass detection from test images, including healthy ovaries and masses (internal test area under the curve [AUC]: 0.970; external test AUC: 0.877). In terms of classification diagnosis prediction, OvaMTA-Diagnosis demonstrated high performance on image-based internal (AUC: 0.941) and external test sets (AUC: 0.941). In video-based external testing, OvaMTA recognised 159 videos with ovarian masses with AUC of 0.911, and is comparable to the performance of senior radiologists (ACC: 86.2 vs. 88.1, p = 0.50; SEN: 81.8 vs. 88.6, p = 0.16; SPE: 89.2 vs. 87.6, p = 0.68). There was a significant improvement in junior and intermediate radiologists who were assisted by AI compared to those who were not assisted by AI (ACC: 80.8 vs. 75.3, p = 0.00015; SEN: 79.5 vs. 74.6, p = 0.029; SPE: 81.7 vs. 75.8, p = 0.0032). General practitioners assisted by AI achieved an average performance of radiologists (ACC: 82.7 vs. 81.8, p = 0.80; SEN: 84.8 vs. 82.6, p = 0.72; SPE: 81.2 vs. 81.2, p > 0.99).

Interpretation

The OvaMTA system based on ultrasound imaging is a simple and practical auxiliary tool for screening for ovarian cancer, with a diagnostic performance comparable to that of senior radiologists. This provides a potential tool for screening ovarian cancer.

Funding

This work was supported by the National Natural Science Foundation of China (Grant Nos. 12090020, 82071929, and 12090025) and the R&D project of the Pazhou Lab (Huangpu) (Grant No. 2023K0605).

Exploratory study on the enhancement of O-RADS application effectiveness for novice ultrasonographers via deep learning

PURPOSE

The study aimed to create a deep convolutional neural network (DCNN) model based on ConvNeXt-Tiny to identify classic benign lesions (CBL) from other lesions (OL) within the Ovarian-Adnexal Reporting and Data System (O-RADS), enhancing the system's utility for novice ultrasonographers.

METHODS

Two sets of sonographic images of pathologically confirmed adnexal lesions were retrospectively collected [development dataset (DD) and independent test dataset (ITD)]. The ConvNeXt-Tiny model, optimized through transfer learning, was trained on the DD using the original images directly and after automatic lesion segmentation by a U-Net model. Models derived from both training paradigms were validated on the ITD for sensitivity, specificity, accuracy, and area under the curve (AUC). Two novice ultrasonographers were assessed in O-RADS with and without assistance from the model for Application Effectiveness.

RESULTS

The ConvNeXt-Tiny model trained on original images scored AUCs of 0.978 for DD and 0.955 for ITD, while the U-Net segmented image model achieved 0.967 for DD and 0.923 for ITD; neither showed significant differences. When assessing the malignancy of lesions using O-RADS 4 and 5, the diagnostic performances of two novice ultrasonographers and senior ultrasonographer, as well as model-assisted classifications, showed no significant differences, except for one novice's low accuracy. This approach reduced classification time by 62 and 64 min. The kappa values with senior doctors' classifications rose from 0.776 and 0.761 to 0.914 and 0.903, respectively.

CONCLUSION

The ConvNeXt-Tiny model demonstrated excellent and stable performance in distinguishing CBL from OL within O-RADS. The diagnostic performance of novice ultrasonographers using O-RADS is essentially equivalent to that of senior ultrasonographer, and the assistance of the model can enhance their classification efficiency and consistency with the results of senior ultrasonographer.

Reliability, reproducibility, and potential pitfalls of the O-RADS scoring with non-dynamic MRI

BACKGROUND

The O-RADS scoring has been proposed to standardize the reporting of adnexal lesions using magnetic resonance imaging (MRI).

PURPOSE

To assess intra- and inter-observer agreement of the O-RADS scoring using non-dynamic MRI and its agreement with pathologic diagnosis, and to provide the pitfalls in the scoring based on discordant ratings.

MATERIAL AND METHODS

Adnexal lesions that were diagnosed using non-dynamic MRI at two centers were scored using O-RADS. Intra- and inter-observer agreements were assessed using kappa statistics. Cross-tabulations were made for intra- and inter-observer ratings and for O-RADS scores and pathological findings.

RESULTS

Intra- and inter-observer agreements were assessed for 404 lesions in 339 patients who were admitted to center 1. Intra-observer agreement was almost perfect (97.8%, kappa = 0.963) and inter-observer agreement was substantial (83.2%, kappa = 0.730). The combined data from center 1 and center 2 included 496 patients; of them, 295 (59.5%) were operated. There was no borderline or malignant pathology for the lesions with O-RADS 1 or 2. Of those with an O-RADS score of 3, 3 (4.1%) lesions were borderline and none were malignant. The O-RADS scoring in discriminating borderline/malignant lesions from benign lesions was outstanding (area under the ROC curve 0.950, 95% CI = 0.923-0.971). Sensitivity, specificity, positive, and negative predictive values of O-RADS 4/5 lesions for borderline/malignant lesions were 96.2%, 87.1%, 72.8%, and 98.4%, respectively.

CONCLUSION

The O-RADS scoring using non-dynamic MRI is a reproducible method and has good discrimination for borderline/malignant lesions. Potential factors that may lead to discordant ratings are provided here.

Clinical Utility and Diagnostic Accuracy of ROMA, RMI, ADNEX, HE4, and CA125 in the Prediction of Malignancy in Adnexal Masses

OBJECTIVE

We aimed to compare the clinical utility and diagnostic accuracy of the ADNEX model, ROMA score, RMI I, and RMI IV, as well as two serum markers (CA125 and HE4) in preoperative discrimination between benign and malignant adnexal masses (AMs).

METHODS

We conducted a retrospective study extracting all consecutive patients with AMs seen at our Institution between January 2015 and December 2020. Accuracy metrics included sensitivity (SE), specificity (SP), and area under the receiver operating characteristic curve (AUC), and their 95% confidence intervals (CI) were calculated for basic discrimination between AMs. Model performance was evaluated in terms of discrimination ability and clinical utility (net benefit, NB).

RESULTS

A total of 581 women were included; 481 (82.8%) had a benign ovarian tumor and 100 (17.2%) had a malignant tumor. The SE and SP of CA125, HE4, ROMA score, RMI I, RMI IV, and ADNEX model were 0.60 (0.54-0.66) and 0.80 (0.76-0.83); 0.39 (0.30-0.49) and 0.96 (0.94-0.98); 0.59 (0.50-0.68) and 0.92 (0.88-0.95); 0.56 (0.46-0.65) and 0.98 (0.96-0.99); 0.54 (0.44-0.63) and 0.96 (0.94-0.98); 0.82 (0.73-0.88) and 0.91 (0.89-0.94), respectively. The overall AUC was 0.76 (0.74-0.79) for CA125, 0.81 (0.78-0.83) for HE4, 0.82 (0.80-0.85) for ROMA, 0.86 (0.84-0.88) for RMI I, 0.83 (0.81-0.86) for RMI IV, and 0.92 (0.90-0.94) for ADNEX. The NB for ADNEX was higher than other biomarkers and models across all decision thresholds between 5% and 50%.

CONCLUSIONS

The ADNEX model showed a better discrimination ability and clinical utility when differentiating malignant from benign Ams, compared to CA125, HE4, ROMA score, RMI I, and RMI IV.

Machine learning models in evaluating the malignancy risk of ovarian tumors: a comparative study

OBJECTIVES

The study aimed to compare the diagnostic efficacy of the machine learning models with expert subjective assessment (SA) in assessing the malignancy risk of ovarian tumors using transvaginal ultrasound (TVUS).

METHODS

The retrospective single-center diagnostic study included 1555 consecutive patients from January 2019 to May 2021. Using this dataset, Residual Network(ResNet), Densely Connected Convolutional Network(DenseNet), Vision Transformer(ViT), and Swin Transformer models were established and evaluated separately or combined with Cancer antigen 125 (CA 125). The diagnostic performance was then compared with SA.

RESULTS

Of the 1555 patients, 76.9% were benign, while 23.1% were malignant (including borderline). When differentiating the malignant from ovarian tumors, the SA had an AUC of 0.97 (95% CI, 0.93-0.99), sensitivity of 87.2%, and specificity of 98.4%. Except for Vision Transformer, other machine learning models had diagnostic performance comparable to that of the expert. The DenseNet model had an AUC of 0.91 (95% CI, 0.86-0.95), sensitivity of 84.6%, and specificity of 95.1%. The ResNet50 model had an AUC of 0.91 (0.85-0.95). The Swin Transformer model had an AUC of 0.92 (0.87-0.96), sensitivity of 87.2%, and specificity of 94.3%. There was a statistically significant difference between the Vision Transformer and SA, and between the Vision Transformer and Swin Transformer models (AUC: 0.87 vs. 0.97, P = 0.01; AUC: 0.87 vs. 0.92, P = 0.04). Adding CA125 did not improve the diagnostic performance of the models in distinguishing benign and malignant ovarian tumors.

CONCLUSION

The deep learning model of TVUS can be used in ovarian cancer evaluation, and its diagnostic performance is comparable to that of expert assessment.

Do we more often opt for conservative management of ovarian tumors after changing the Dutch national guideline on enlarged ovaries? A nationwide cohort study

INTRODUCTION

Increasing evidence shows that conservative management of ovarian tumors classified as benign, based on ultrasound assessment, is safe. Therefore, conservative management has been adopted as the preferred strategy for certain ovarian tumors assessed as benign in the Dutch national guideline on enlarged ovaries in 2013. The aim of this study was to examine whether implementation of this guideline has led to changes in the number of women/100 000 women undergoing surgery for an ovarian tumor in the Netherlands.

MATERIAL AND METHODS

Histopathology reports were requested for all examinations of ovarian and fallopian tube specimens (including cyst enucleations) registered in Palga, the Dutch nationwide pathology databank, from 2011 (before guideline adaptation) and 2019 (after guideline adaptation). Reports on prophylactically removed adnexa, removal for other primary tumors (eg endometrial carcinoma), and for patients under 18 years of age, were excluded from the analysis. Interobserver agreement for the inclusion and classification of reports was assessed using Cohen's Kappa analysis.

RESULTS

A total of 34 932 reports were retrieved, 13 917 of which were included in the analysis. In 2011 and 2019, respectively, 96.3/100 000 vs 68.8/100 000 women aged ≥18 underwent surgery for benign ovarian tumors, and 19.6/100 000 vs 18.3/100 000 for borderline and malignant tumors combined. The number of women/100 000 who had surgery for a benign ovarian tumor per 100 000 women declined by 28.5% (p < 0.001) between 2011 and 2019. The largest difference between 2011 and 2019 was observed in the number of women per 100 000 women who underwent surgery for a serous cystadenoma (-40.7%; 20.8/100 000 vs. 12.3/100 000), followed by endometrioma (-33.2%; 14.7/100 000 vs. 9.8/100 000), simple epithelial cyst (-57.3%; 8.4/100 000 vs. 3.6/100 000), and corpus luteum cyst (-57.0%; 4.0/100 000 vs. 1.7/100 000). Cohen's Kappa for the interobserver agreement was 0.96.

CONCLUSIONS

The number of women/100 000 undergoing surgery for a benign ovarian tumor has substantially decreased in the Netherlands when comparing data before and after implementation of the national guideline in 2013, while the number of women/100 000 undergoing surgery for a malignant or borderline tumor remained the same. These findings suggest successful implementation of the updated guideline, and a measurable effect on increased adoption of conservative management for benign-looking ovarian tumors.

The educational game SonoQz improves diagnostic performance in ultrasound assessment of ovarian tumors

INTRODUCTION

Our objective was to determine whether the educational game SonoQz can improve diagnostic performance in ultrasound assessment of ovarian tumors.

MATERIAL AND METHODS

The SonoQz mobile application was developed as an educational tool for medical doctors to practice ultrasound assessment, based on still images of ovarian tumors. The game comprises images from 324 ovarian tumors, examined by an ultrasound expert prior to surgery. A training phase, where the participants assessed at least 200 cases in the SonoQz app, was preceded by a pretraining test, and followed by a posttraining test. Two equal tests (A and B), each consisting of 20 cases, were used as pre- and posttraining tests. Half the users took test A first, B second, and the remaining took the tests in the opposite order. Users were asked to classify the tumors (1) according to International Ovarian Tumor Analysis (IOTA) Simple Rules, (2) as benign or malignant, and (3) suggest a specific histological diagnosis. Logistic mixed models with fixed effects for pre- and posttraining tests, and crossed random effects for participants and cases, were used to determine any improvement in test scores, sensitivity, and specificity.

RESULTS

Fifty-eight doctors from 19 medical centers participated. Comparing the pre- and posttraining test, the median of correctly classified cases, in Simple Rules assessment increased from 72% to 83%, p < 0.001; in classifying the lesion as benign or malignant tumors from 86% to 95%, p < 0.001; and in making a specific diagnosis from 43% to 63%, p < 0.001. When classifying tumors as benign or malignant, at an unchanged level of sensitivity (98% vs. 97%, p = 0.157), the specificity increased from 70% to 89%, p < 0.001.

CONCLUSIONS

Our results indicate that the educational game SonoQz is an effective tool that may improve diagnostic performance in assessing ovarian tumors, specifically by reducing the number of false positives while maintaining high sensitivity.

Ovarian cancer risk among older patients with stable adnexal masses

BACKGROUND

Few studies have evaluated the risk of cancer among older patients with stable adnexal masses in community-based settings to determine the duration of observation time needed.

OBJECTIVE

This study aimed to assess the ovarian cancer risk among older patients with stable adnexal masses on ultrasound.

STUDY DESIGN

This was a retrospective cohort study of patients in a large community-based health system aged ≥50 years with an adnexal mass <10 cm on ultrasound between 2016 and 2020 who had at least 1 follow-up ultrasound performed ≥6 weeks after initial ultrasound. Masses were considered stable on follow-up examination if they did not exhibit an increase of >1 cm in the greatest dimension or a change in standardized reported ultrasound characteristics. Ovarian cancer risk was determined at increasing time intervals of stability after initial ultrasound.

RESULTS

Among 4061 patients with stable masses, the average age was 61 years (range, 50-99), with an initial mass size of 3.8 cm (range, 0.2-9.9). With a median follow-up of 3.7 years, 11 cancers were detected, with an absolute risk of 0.27%. Ovarian cancer risk declined with longer duration of stability, from 0.73 (95% confidence interval, 0.30-1.17) per 1000 person-years at 6 to 12 weeks, 0.63 (95% confidence interval, 0.19-1.07) at 13 to 24 weeks, 0.44 (95% confidence interval, 0.01-0.87) at 25 to 52 weeks, and 0.00 (95% confidence interval, 0.00-0.00) at >52 weeks. Expressed as number needed to reimage, ongoing ultrasound imaging would be needed for 369 patients whose masses show stability at 6 to 12 weeks, 410 patients at 13 to 24 weeks, 583 patients at 25 to 52 weeks, and >1142 patients with stable masses at 53 to 104 weeks to detect 1 case of ovarian cancer.

CONCLUSION

In a diverse community-based setting, among patients aged ≥50 years with an adnexal mass that was stable for at least 6 weeks after initial ultrasound, the risk of ovarian cancer was very low at 0.27%. Longer demonstrated duration of stability was associated with progressively lower risk, with no cancer cases observed after 52 weeks of stability. These findings suggest that the benefit of ultrasound monitoring of stable masses beyond 12 months is minimal and may be outweighed by potential risks of repeated imaging.

Validation of ADNEX and IOTA two-step strategy and estimation of risk of complications during follow-up of adnexal masses in low-risk population

OBJECTIVES

To evaluate the ability of the Assessment of Different NEoplasias in the adneXa (ADNEX) model and the International Ovarian Tumour Analysis (IOTA) two-step strategy to predict malignancy in adnexal masses detected in an outpatient low-risk setting, and to estimate the risk of complications in masses with benign ultrasound morphology managed using clinical and ultrasound follow-up.

METHODS

This single-center study was performed at Hospital Universitari Dexeus, Barcelona, Spain, using interim data from the ongoing prospective observational IOTA Phase-5 (IOTA5) study. The primary aim of the IOTA5 study is to describe the cumulative incidence of complications during follow-up of adnexal masses classified as benign on ultrasound examination. Consecutive patients with an adnexal mass detected between June 2012 and September 2016 in a private center offering screening for gynecological cancer were included and followed up until February 2020. Tumors were classified as benign or malignant based on histology (if patients underwent surgery) or the outcome of clinical and ultrasound follow-up at 12 (range, 10-14) months. Multiple imputation was used when outcomes were uncertain. The ability of the ADNEX model without CA125 and of the IOTA two-step strategy to distinguish benign from malignant masses was evaluated retrospectively using the prospectively collected data. We assessed performance with regard to discrimination (area under the receiver-operating-characteristics curve (AUC)), calibration, classification (sensitivity and specificity) and clinical utility (Net Benefit). In the group of patients with a mass judged to be benign who were selected for conservative management, we evaluated the occurrence of spontaneous resolution or any mass complication during the first 5 years of follow-up by assessing the cumulative incidence of malignancy, torsion, cyst rupture and minor mass complications (inflammation, infection or adhesions) and the time to occurrence of an event.

RESULTS

A total of 2654 patients were recruited to the study. After application of exclusion criteria, 2039 patients with a newly detected mass were included for the model validation. Of those, 1684 (83%) masses were benign, 49 (2%) masses were malignant and, for 306 (15%) masses, the outcome was uncertain and therefore imputed. The AUC was 0.95 (95% CI, 0.89-0.98) for ADNEX without CA125 and 0.94 (95% CI, 0.88-0.97) for the two-step strategy. Calibration performance could not be meaningfully interpreted because the small number of malignancies resulted in very wide confidence intervals. The two-step strategy had better clinical utility than did the ADNEX model at malignancy risk thresholds < 3%. There were 1472 (72%) patients whose mass was judged to be benign based on pattern recognition by an experienced ultrasound examiner and were managed with clinical and ultrasound follow-up. In this group, the 5-year cumulative incidence was 66% (95% CI, 63-69%) for spontaneous resolution of the mass, 0% (95% CI, 0-0.2%) for torsion, 0.1% (95% CI, < 0.1-0.4%) for cyst rupture, 0.2% (95% CI, 0.1-0.6%) for a borderline tumor and 0.2% (95% CI, 0.1-0.6%) for invasive malignancy.

CONCLUSIONS

The ADNEX model and IOTA two-step strategy performed well to distinguish benign from malignant adnexal masses detected in a low-risk population. Conservative management is safe for masses with a benign ultrasound appearance in this population. © 2024 International Society of Ultrasound in Obstetrics and Gynecology.

Clinical value of ACR O-RADS combined with CA125 in the risk stratification of adnexal masses

Purpose

To develop a combined diagnostic model integrating the subclassification of the 2022 version of the American College of Radiology (ACR) Ovarian-Adnexal Reporting and Data System (O-RADS) with carbohydrate antigen 125 (CA125) and to validate whether the combined model can offer superior diagnostic efficacy than O-RADS alone in assessing adnexal malignancy risk.

Methods

A retrospective analysis was performed on 593 patients with adnexal masses (AMs), and the pathological and clinical data were included. According to the large differences in malignancy risk indices for different image features in O-RADS category 4, the lesions were categorized into groups A and B. A new diagnostic criterion was developed. Lesions identified as category 1, 2, 3, or 4A with a CA125 level below 35 U/ml were classified as benign. Lesions identified as category 4A with a CA125 level more than or equal to 35 U/ml and lesions with a category of 4B and 5 were classified as malignant. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and area under the curve (AUC) of O-RADS (v2022), CA125, and the combined model in the diagnosis of AMs were calculated and compared.

Results

The sensitivity, specificity, PPV, NPV, accuracy, and AUCs of the combined model were 92.4%, 96.5%, 80.2%, 98.8%, 94.1%, and 0.945, respectively. The specificity, PPV, accuracy, and AUC of the combined model were significantly higher than those of O-RADS alone (all P < 0.01). In addition, both models had acceptable sensitivity and NPV, but there were no significant differences among them (P > 0.05).

Conclusion

The combined model integrating O-RADS subclassification with CA125 could improve the specificity and PPV in diagnosing malignant AMs. It could be a valuable tool in the clinical application of risk stratification of AMs.

Neural network-derived multivariate index assay demonstrates effective clinical performance in longitudinal monitoring of ovarian cancer risk

OBJECTIVE

We recently characterized the clinical performance of a multivariate index assay (MIA3G) to assess ovarian cancer risk for adnexal masses at initial presentation. This study evaluated how MIA3G varies when applied longitudinally to monitor risk during clinical follow-up.

METHOD

The study evaluated women presenting with adnexal masses from eleven centers across the US. Patients received an initial blood draw at enrollment and at the standard-of-care follow-up visits. MIA3G was determined for all visits but physicians did not have access to MIA3G scores to determine clinical management. The primary outcome was the relative change value (RCV) of MIA3G over the period of clinical observation.

RESULTS

A total of 510 patients of 785 enrolled met study criteria. Of these, 30.8% had a second, 25.4% a third and 22.2% a fourth blood draw following initial collection. The median duration from initial draw was 131 d to second draw, 301.5 d to the third draw and 365.5 d to the fourth draw. MIA3G RCV of >50% was observed in 22-26% patients, whereas 70-75% patients had MIA3G RCV >5%. An empirical baseline RCV of 56% - transformed to 1 in logarithmic scale - was calculated from averaging RCVs of all patients who had no malignancy risk after 210 days. RCV > 1 log was associated with higher incidence of surgical intervention (29.6%) compared to RCV < 1 log (16.9%).

CONCLUSIONS

Variation in MI3AG does not change the accuracy of the test for excluding malignancy, while marked changes may be associated with a slightly higher likelihood of surgical intervention. In addition to MIA3G score itself, the MIA3G RCV may be important for clinical management.

ACR Appropriateness Criteria® Clinically Suspected Adnexal Mass, No Acute Symptoms: 2023 Update

Asymptomatic adnexal masses are commonly encountered in daily radiology practice. Although the vast majority of these masses are benign, a small subset have a risk of malignancy, which require gynecologic oncology referral for best treatment outcomes. Ultrasound, using a combination of both transabdominal, transvaginal, and duplex Doppler technique can accurately characterize the majority of these lesions. MRI with and without contrast is a useful complementary modality that can help characterize indeterminate lesions and assess the risk of malignancy is those that are suspicious. The American College of Radiology Appropriateness Criteria are evidence-based guidelines for specific clinical conditions that are reviewed annually by a multidisciplinary expert panel. The guideline development and revision process support the systematic analysis of the medical literature from peer reviewed journals. Established methodology principles such as Grading of Recommendations Assessment, Development, and Evaluation or GRADE are adapted to evaluate the evidence. The RAND/UCLA Appropriateness Method User Manual provides the methodology to determine the appropriateness of imaging and treatment procedures for specific clinical scenarios. In those instances where peer reviewed literature is lacking or equivocal, experts may be the primary evidentiary source available to formulate a recommendation.

An explainable machine learning model to solid adnexal masses diagnosis based on clinical data and qualitative ultrasound indicators

BACKGROUND

Accurate characterization of newly diagnosed a solid adnexal lesion is a key step in defining the most appropriate therapeutic approach. Despite guidance from the International Ovarian Tumor Analyzes Panel, the evaluation of these lesions can be challenging. Recent studies have demonstrated how machine learning techniques can be applied to clinical data to solve this diagnostic problem. However, ML models can often consider as black-boxes due to the difficulty of understanding the decision-making process used by the algorithm to obtain a specific result.

AIMS

For this purpose, we propose an Explainable Artificial Intelligence model trained on clinical characteristics and qualitative ultrasound indicators to predict solid adnexal masses diagnosis.

MATERIALS & METHODS

Since the diagnostic task was a three-class problem (benign tumor, invasive cancer, or ovarian metastasis), we proposed a waterfall classification model: a first model was trained and validated to discriminate benign versus malignant, a second model was trained to distinguish nonmetastatic versus metastatic malignant lesion which occurs when a patient is predicted to be malignant by the first model. Firstly, a stepwise feature selection procedure was implemented. The classification performances were validated on Leave One Out scheme.

RESULTS

The accuracy of the three-class model reaches an overall accuracy of 86.36%, and the precision per-class of the benign, nonmetastatic malignant, and metastatic malignant classes were 86.96%, 87.27%, and 77.78%, respectively.

DISCUSSION

SHapley Additive exPlanations were performed to visually show how the machine learning model made a specific decision. For each patient, the SHAP values expressed how each characteristic contributed to the classification result. Such information represents an added value for the clinical usability of a diagnostic system.

CONCLUSIONS

This is the first work that attempts to design an explainable machine-learning tool for the histological diagnosis of solid masses of the ovary.

Utility of a Multi-Marker Panel with Ultrasound for Enhanced Classification of Adnexal Mass

Pre-surgical clinical assessment of an adnexal mass typically relies on transvaginal ultrasound for comprehensive morphological assessment, with further support provided by biomarker measurements and clinical evaluation. Whilst effective for masses that are obviously benign or malignant, a large proportion of masses remain sonographically indeterminate at surgical referral. As a consequence, post-surgical diagnoses of benign disease can outnumber malignancies up to 9-fold, while less than 50% of cancer cases receive a primary referral to a gynecological oncology specialist. We recently described a blood biomarker signature (multi-marker panel-MMP) that differentiated patients with benign from malignant ovarian disease with high accuracy. In this study, we have examined the use of the MMP, both individually and in combination with transvaginal ultrasound, as an alternative tool to CA-125 for enhanced decision making in the pre-surgical referral process.

Ultrasound image-based nomogram combining clinical, radiomics, and deep transfer learning features for automatic classification of ovarian masses according to O-RADS

Background

Accurate and rapid discrimination between benign and malignant ovarian masses is crucial for optimal patient management. This study aimed to establish an ultrasound image-based nomogram combining clinical, radiomics, and deep transfer learning features to automatically classify the ovarian masses into low risk and intermediate-high risk of malignancy lesions according to the Ovarian- Adnexal Reporting and Data System (O-RADS).

Methods

The ultrasound images of 1,080 patients with 1,080 ovarian masses were included. The training cohort consisting of 683 patients was collected at the South China Hospital of Shenzhen University, and the test cohort consisting of 397 patients was collected at the Shenzhen University General Hospital. The workflow included image segmentation, feature extraction, feature selection, and model construction.

Results

The pre-trained Resnet-101 model achieved the best performance. Among the different mono-modal features and fusion feature models, nomogram achieved the highest level of diagnostic performance (AUC: 0.930, accuracy: 84.9%, sensitivity: 93.5%, specificity: 81.7%, PPV: 65.4%, NPV: 97.1%, precision: 65.4%). The diagnostic indices of the nomogram were higher than those of junior radiologists, and the diagnostic indices of junior radiologists significantly improved with the assistance of the model. The calibration curves showed good agreement between the prediction of nomogram and actual classification of ovarian masses. The decision curve analysis showed that the nomogram was clinically useful.

Conclusion

This model exhibited a satisfactory diagnostic performance compared to junior radiologists. It has the potential to improve the level of expertise of junior radiologists and provide a fast and effective method for ovarian cancer screening.

Ovarian Cancer surgical consideration is markedly improved by the neural network powered-MIA3G multivariate index assay

Background

Surgery remains the main treatment option for an adnexal mass suspicious of ovarian cancer. The malignancy rate is, however, only 10-15% in women undergoing surgery. This results in a high number of unnecessary surgeries. A surveillance-based approach is recommended to form the basis for surgical referrals. We have previously reported the clinical performance of MIA3G, a deep neural network-based algorithm, for assessing ovarian cancer risk. In this study, we show that MIA3G markedly improves the surgical selection for women presenting with adnexal masses.

Methods

MIA3G employs seven serum biomarkers, patient age, and menopausal status. Serum samples were collected from 785 women (IQR: 39-55 years) across 12 centers that presented with adnexal masses. MIA3G risk scores were calculated for all subjects in this cohort. Physicians had no access to the MIA3G risk score when deciding upon a surgical referral. The performance of MIA3G for surgery referral was compared to clinical and surgical outcomes. MIA3G was also tested in an independent cohort comprising 29 women across 14 study sites, in which the physicians had access to and utilized MIA3G prior to surgical consideration.

Results

When compared to the actual number of surgeries (n = 207), referrals based on the MIA3G score would have reduced surgeries by 62% (n = 79). The reduction was higher in premenopausal patients (77%) and in patients ≤55 years old (70%). In addition, a 431% improvement in malignancy prediction would have been observed if physicians had utilized MIA3G scores for surgery selection. The accuracy of MIA3G referral was 90.00% (CI 87.89-92.11), while only 9.18% accuracy was observed when the MIA3G score was not used. These results were corroborated in an independent multi-site study of 29 patients in which the physicians utilized MIA3G in surgical consideration. The surgery reduction was 87% in this cohort. Moreover, the accuracy and concordance of MIA3G in this independent cohort were each 96.55%.

Conclusion

These findings demonstrate that MIA3G markedly augments the physician's decisions for surgical intervention and improves malignancy prediction in women presenting with adnexal masses. MIA3G utilization as a clinical diagnostic tool might help reduce unnecessary surgeries.

Multi-modality deep learning model reaches high prediction accuracy in the diagnosis of ovarian cancer

We evaluated the diagnostic performance of a multimodal deep-learning (DL) model for ovarian mass differential diagnosis. This single-center retrospective study included 1,054 ultrasound (US)-detected ovarian tumors (699 benign and 355 malignant). Patients were randomly divided into training (n = 675), validation (n = 169), and testing (n = 210) sets. The model was developed using ResNet-50. Three DL-based models were proposed for benign-malignant classification of these lesions: single-modality model that only utilized US images; dual-modality model that used US images and menopausal status as inputs; and multi-modality model that integrated US images, menopausal status, and serum indicators. After 5-fold cross-validation, 210 lesions were tested. We evaluated the three models using the area under the curve (AUC), accuracy, sensitivity, and specificity. The multimodal model outperformed the single- and dual-modality models with 93.80% accuracy and 0.983 AUC. The Multimodal ResNet-50 DL model outperformed the single- and dual-modality models in identifying benign and malignant ovarian tumors.

O-RADS MRI scoring system: key points for correct application in inexperienced hands

OBJECTIVES

To evaluate the efficacy of the O-RADS MRI criteria in the stratification of risk of malignancy of solid or sonographically indeterminate ovarian masses and assess the interobserver agreement of this classification between experienced and inexperienced radiologists.

METHODS

This single-centre retrospective study included patients from 2019 to 2022 with sonographically indeterminate or solid ovarian masses who underwent MRI with a specific protocol for characterisation according to O-RADS MRI specifications. Each study was evaluated using O-RADS lexicon by two radiologists, one with 17 years of experience in gynaecological radiology and another with 4 years of experience in general radiology. Findings were classified as benign, borderline, or malignant according to histology or stability over time. Diagnostic performance and interobserver agreement were assessed.

RESULTS

A total of 183 patients with US indeterminate or solid adnexal masses were included. Fifty-seven (31%) did not have ovarian masses, classified as O-RADS 1. The diagnostic performance for scores 2-5 was excellent with a sensitivity, specificity, PPV, and NPV of 97.4%, 100%, 96.2%, and 100%, respectively by the experienced radiologist and 96.1%, 92.0%, 93.9%, and 94.8% by the inexperienced radiologist. Interobserver concordance was very high (Kappa index 0.92). Almost all the misclassified cases were due to misinterpretation of the classification similar to reports in the literature.

CONCLUSION

The diagnostic performance of O-RADS MRI determined by either experienced or inexperienced radiologists is excellent, facilitating decision-making with high diagnostic accuracy and high reproducibility. Knowledge of this classification and use of assessment tools could avoid frequent errors due to misinterpretation.

CRITICAL RELEVANCE STATEMENT

Up to 31% of ovarian masses are considered indeterminate by transvaginal US and 32% of solid lesions considered malignant by transvaginal US are benign. The O-RADs MRI accurately classifies these masses, even when used by inexperienced radiologists, thereby avoiding incorrect surgical approaches.

KEY POINTS

• O-RADS MRI accurately classifies indeterminate and solid ovarian masses by ultrasound. • There is excellent interobserver agreement between experienced and non-experienced radiologists. • O-RADS MRI is a helpful tool to assess clinical decision-making in ovarian tumours.

Imaging in gynecological disease (27): clinical and ultrasound characteristics of recurrent ovarian stromal cell tumors

OBJECTIVE

To describe the clinical and ultrasound characteristics of recurrent granulosa cell and Sertoli-Leydig cell tumors.

METHODS

This was a retrospective observational study performed at Fondazione Policlinico Universitario A. Gemelli, Istituto di Ricovero e Cura a Carattere Scientifico, IRCCS, Rome (Gemelli center), Italy. Patients with a histological diagnosis of recurrent granulosa cell tumor or Sertoli-Leydig cell tumor were identified from the database of the Department of Gynecological Oncology. Those who had undergone a preoperative ultrasound examination at the Gemelli center between 2012 and 2020 were included, and the data retrieved from the original ultrasound reports. In all of these reports, the recurrent tumors were described using International Ovarian Tumor Analysis (IOTA) terminology. If a patient had more than one episode of relapse, information from all episodes was collected. If there was more than one recurrent tumor at the same ultrasound examination, all tumors were included. One expert sonographer also reviewed all available ultrasound images to identify typical ultrasound patterns using pattern recognition.

RESULTS

We identified 30 patients with a histological diagnosis of recurrent granulosa cell tumor (25 patients, 55 tumors) or Sertoli-Leydig cell tumor (five patients, seven tumors). All 30 had undergone at least one preoperative ultrasound examination at the Gemelli center and were included. These women had a total of 66 episodes of relapse, of which a preoperative ultrasound examination had been performed at the Gemelli center in 34, revealing 62 recurrent lesions: one in 22/34 (64.7%) episodes of relapse, two in 4/34 (11.8%) episodes and three or more in 8/34 (23.5%) episodes. Most recurrent granulosa cell tumors (38/55, 69.1%) and recurrent Sertoli-Leydig tumors (6/7, 85.7%) were classified as solid or multilocular-solid tumors, while 8/55 (14.5%) recurrent granulosa cell tumors and 1/7 (14.3%) recurrent Sertoli-Leydig cell tumors were unilocular cysts and 9/55 (16.4%) recurrent granulosa cell tumors were multilocular cysts. The nine unilocular cysts had contents that were anechoic (n = 2) or had low-level echogenicity (n = 7), had either smooth (n = 4) or irregular (n = 5) internal cyst walls, and ranged in largest diameter from 8 to 38 mm, with three being < 20 mm and five being 20-30 mm. On retrospective review of the images, two typical ultrasound patterns were described: small solid tumor measuring < 2 cm (15/62, 24.2%) and tumor with vascularized echogenic ground-glass-like content (12/62, 19.4%).

CONCLUSIONS

Some granulosa cell and Sertoli-Leydig cell recurrences manifest one of two typical ultrasound patterns, while some appear as unilocular cysts. These are usually classified as benign, but in patients being followed up for a granulosa cell tumor or Sertoli-Leydig cell tumor, a unilocular cyst should be considered suspicious of recurrence. © 2023 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.

Machine learning and radiomics for segmentation and classification of adnexal masses on ultrasound

Ultrasound-based models exist to support the classification of adnexal masses but are subjective and rely upon ultrasound expertise. We aimed to develop an end-to-end machine learning (ML) model capable of automating the classification of adnexal masses. In this retrospective study, transvaginal ultrasound scan images with linked diagnoses (ultrasound subjective assessment or histology) were extracted and segmented from Imperial College Healthcare, UK (ICH development dataset; n = 577 masses; 1444 images) and Morgagni-Pierantoni Hospital, Italy (MPH external dataset; n = 184 masses; 476 images). A segmentation and classification model was developed using convolutional neural networks and traditional radiomics features. Dice surface coefficient (DICE) was used to measure segmentation performance and area under the ROC curve (AUC), F1-score and recall for classification performance. The ICH and MPH datasets had a median age of 45 (IQR 35-60) and 48 (IQR 38-57) years old and consisted of 23.1% and 31.5% malignant cases, respectively. The best segmentation model achieved a DICE score of 0.85 ± 0.01, 0.88 ± 0.01 and 0.85 ± 0.01 in the ICH training, ICH validation and MPH test sets. The best classification model achieved a recall of 1.00 and F1-score of 0.88 (AUC:0.93), 0.94 (AUC:0.89) and 0.83 (AUC:0.90) in the ICH training, ICH validation and MPH test sets, respectively. We have developed an end-to-end radiomics-based model capable of adnexal mass segmentation and classification, with a comparable predictive performance (AUC 0.90) to the published performance of expert subjective assessment (gold standard), and current risk models. Further prospective evaluation of the classification performance of this ML model against existing methods is required.

ADNEX risk prediction model for diagnosis of ovarian cancer: systematic review and meta-analysis of external validation studies

Objectives

To conduct a systematic review of studies externally validating the ADNEX (Assessment of Different Neoplasias in the adnexa) model for diagnosis of ovarian cancer and to present a meta-analysis of its performance.

Design

Systematic review and meta-analysis of external validation studies.

Data sources

Medline, Embase, Web of Science, Scopus, and Europe PMC, from 15 October 2014 to 15 May 2023.

Eligibility criteria for selecting studies

All external validation studies of the performance of ADNEX, with any study design and any study population of patients with an adnexal mass. Two independent reviewers extracted the data. Disagreements were resolved by discussion. Reporting quality of the studies was scored with the TRIPOD (Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis) reporting guideline, and methodological conduct and risk of bias with PROBAST (Prediction model Risk Of Bias Assessment Tool). Random effects meta-analysis of the area under the receiver operating characteristic curve (AUC), sensitivity and specificity at the 10% risk of malignancy threshold, and net benefit and relative utility at the 10% risk of malignancy threshold were performed.

Results

47 studies (17 007 tumours) were included, with a median study sample size of 261 (range 24-4905). On average, 61% of TRIPOD items were reported. Handling of missing data, justification of sample size, and model calibration were rarely described. 91% of validations were at high risk of bias, mainly because of the unexplained exclusion of incomplete cases, small sample size, or no assessment of calibration. The summary AUC to distinguish benign from malignant tumours in patients who underwent surgery was 0.93 (95% confidence interval 0.92 to 0.94, 95% prediction interval 0.85 to 0.98) for ADNEX with the serum biomarker, cancer antigen 125 (CA125), as a predictor (9202 tumours, 43 centres, 18 countries, and 21 studies) and 0.93 (95% confidence interval 0.91 to 0.94, 95% prediction interval 0.85 to 0.98) for ADNEX without CA125 (6309 tumours, 31 centres, 13 countries, and 12 studies). The estimated probability that the model has use clinically in a new centre was 95% (with CA125) and 91% (without CA125). When restricting analysis to studies with a low risk of bias, summary AUC values were 0.93 (with CA125) and 0.91 (without CA125), and estimated probabilities that the model has use clinically were 89% (with CA125) and 87% (without CA125).

Conclusions

The results of the meta-analysis indicated that ADNEX performed well in distinguishing between benign and malignant tumours in populations from different countries and settings, regardless of whether the serum biomarker, CA125, was used as a predictor. A key limitation was that calibration was rarely assessed.

Systematic review registration

PROSPERO CRD42022373182.

Modified MR scoring system for assessment of sonographically indeterminate ovarian and adnexal masses in the absence of dynamic contrast-enhanced

OBJECTIVES

Dynamic contrast-enhanced (DCE) MRI is not available in all imaging centres to investigate adnexal masses. We proposed modified magnetic resonance (MR) scoring system based on an assessment of the enhancement of the solid tissue on early phase postcontrast series and diffusion-weighted imaging (DWI) with apparent diffusion coefficient (ADC) map and investigated the validity of this protocols in the current study.

MATERIALS AND METHODS

In this cross-sectional retrospective study, pelvic MRI of a total of 245 patients with 340 adnexal masses were studied based on the proposed modified scoring system and ADNEX MR scoring system.

RESULTS

Modified scoring system with the sensitivity of 87.3% and specificity of 94.6% has an accuracy of 92.1%. Sensitivity, specificity, and accuracy of ADNEX MR scoring system is 96.6%, 91%, and 92.9%, respectively. The area under the receiver operating characteristic curve for the modified scoring system and ADNEX MR scoring system is 0.909 (with 0.870-0.938 95% confidence interval [CI]) and 0.938 (with 0.907-0.961 95% CI), respectively. Pairwise comparison of these area under the curves showed no significant difference (P = .053).

CONCLUSIONS

Modified scoring system is less sensitive than the ADNEX MR scoring system and more specific but the accuracy is not significantly different.

ADVANCES IN KNOWLEDGE

According to our study, MR scoring system based on subjective assessment of the enhancement of the solid tissue on early phase postcontrast series and DWI with ADC map could be applicable in imaging centres that DCE is not available.

Application of O-RADS US combined with MV-Flow to diagnose ovarian-adnexal tumors

PURPOSE

This study aimed to explore the application of Ovarian-Adnexal Reporting and Data System Ultrasound (O-RADS US) combined with MV-Flow (Samsung Medison Co., Ltd.) to diagnose ovarian-adnexal masses.

METHODS

A total of 112 ovarian-adnexal masses (81 benign and 31 malignant) from 105 consecutive patients were analyzed. The O-RADS US and vascular index from MV-Flow (VIMV) were measured and compared with the reference standard. O-RADS US and MV-Flow were tested for consistency.

RESULTS

Receiver operating characteristic curves were drawn for O-RADS US, MV-Flow, and their combination. The combined methods had the largest area under the curve (0.955), followed by O-RADS US (0.929) and MV-Flow (0.923). A mass was considered malignant when the O-RADS US classification was 5 and VIMV was ≥7.15. With this definition, MV-Flow had the highest sensitivity (87.10%), with consistent findings for the combined diagnostic methods and O-RADS US (83.87%). The specificity of the combined diagnostic methods (93.83%) was higher than that of MV-Flow (91.36%). O-RADS US had the lowest specificity (90.12%). The combined diagnostic methods had the highest coincidence rate (91.07%), and MV-Flow (90.18%) had a significantly higher coincidence rate than O-RADS US (88.39%). Both O-RADS US and MV-Flow showed good consistency among different physicians (former kappa, 0.974; latter intraclass correlation coefficient [ICC], 0.986). MV-Flow had a high consistency for the same physician (ICC, 1).

CONCLUSION

O-RADS US and MV-Flow exhibited good diagnostic efficacy, and their combined diagnostic efficacy was higher than that of each individually. O-RADS US and MV-Flow can improve the diagnosis of benign and malignant ovarian-adnexal masses.

Multiclass risk models for ovarian malignancy: an illustration of prediction uncertainty due to the choice of algorithm

BACKGROUND

Assessing malignancy risk is important to choose appropriate management of ovarian tumors. We compared six algorithms to estimate the probabilities that an ovarian tumor is benign, borderline malignant, stage I primary invasive, stage II-IV primary invasive, or secondary metastatic.

METHODS

This retrospective cohort study used 5909 patients recruited from 1999 to 2012 for model development, and 3199 patients recruited from 2012 to 2015 for model validation. Patients were recruited at oncology referral or general centers and underwent an ultrasound examination and surgery ≤ 120 days later. We developed models using standard multinomial logistic regression (MLR), Ridge MLR, random forest (RF), XGBoost, neural networks (NN), and support vector machines (SVM). We used nine clinical and ultrasound predictors but developed models with or without CA125.

RESULTS

Most tumors were benign (3980 in development and 1688 in validation data), secondary metastatic tumors were least common (246 and 172). The c-statistic (AUROC) to discriminate benign from any type of malignant tumor ranged from 0.89 to 0.92 for models with CA125, from 0.89 to 0.91 for models without. The multiclass c-statistic ranged from 0.41 (SVM) to 0.55 (XGBoost) for models with CA125, and from 0.42 (SVM) to 0.51 (standard MLR) for models without. Multiclass calibration was best for RF and XGBoost. Estimated probabilities for a benign tumor in the same patient often differed by more than 0.2 (20% points) depending on the model. Net Benefit for diagnosing malignancy was similar for algorithms at the commonly used 10% risk threshold, but was slightly higher for RF at higher thresholds. Comparing models, between 3% (XGBoost vs. NN, with CA125) and 30% (NN vs. SVM, without CA125) of patients fell on opposite sides of the 10% threshold.

CONCLUSION

Although several models had similarly good performance, individual probability estimates varied substantially.

Early Detection of Ovarian Cancer

The risk of death from ovarian cancer is highly associated with the clinical stage at diagnosis. Efforts to implement screening for ovarian cancer have been largely unsuccessful, due to the low prevalence of the disease in the general population and the heterogeneity of the various cancer types that fall under the ovarian cancer designation. A practical test for early detection will require both high sensitivity and high specificity to balance reducing the number of cancer deaths with minimizing surgical interventions for false positive screens. The technology must be cost-effective to deliver at scale, widely accessible, and relatively noninvasive. Most importantly, a successful early detection test must be effective not only at diagnosing ovarian cancer but also in reducing ovarian cancer deaths. Stepwise or multimodal approaches among the various areas under investigation will likely be required to make early detection a reality.

Improving risk stratification of indeterminate adnexal masses on MRI: What imaging features help predict malignancy in O-RADS MRI 4 lesions?

PURPOSE

Ovarian-Adnexal Reporting and Data System (O-RADS) MRI uses a 5-point scale to establish malignancy risk in sonographically-indeterminate adnexal masses. The management of O-RADS MRI score 4 lesions is challenging, as the prevalence of malignancy is widely variable (5-90%). We assessed imaging features that may sub-stratify O-RADS MRI 4 lesions into malignant and benign subgroups.

METHOD

Retrospective single-institution study of women with O-RADS MRI score of 4 adnexal masses between April 2021-August 2022. Imaging findings were assessed independently by 2 radiologists according to the O-RADS lexicon white paper. MRI and clinical findingswere compared between malignant and benign adnexal masses, and inter-reader agreement was calculated.

RESULTS

Seventy-four women (median age 52 years, IQR 36-61) were included. On pathology, 41 (55.4%) adnexal masses were malignant. Patients with malignant masses were younger (p = 0.02) with higher CA-125 levels (p = 0.03). Size of solid tissue was greater in malignant masses (p = 0.01-0.04). Papillary projections and larger solid portion were more common in malignant lesions; irregular septations and predominantly solid composition were more frequent in benign lesions (p < 0.01). Solid tissue of malignant lesions was more often hyperintense on T2-weighted and diffusion-weighted imaging (p ≤ 0.03). Other imaging findings were not significantly different (p = 0.09-0.77). Inter-reader agreement was excellent-good for most features (ICC = 0. 662-0.950; k = 0. 650-0.860).

CONCLUSION

Various MRI and clinical features differed between malignant and benign O-RADS MRI score 4 adnexal masses. O-RADS MRI 4 lesions may be sub-stratified (high vs low risk) based on solid tissue characteristics and CA-125 levels.

Clinical features of patients with previous spontaneous rupture of ovarian endometrioma operated electively: a case-control study

BACKGROUND

The aim of the study is to investigate the proportion and clinical features of previous spontaneously ruptured ovarian endometrioma among women who underwent elective surgery for endometrioma.

METHODS

This retrospective study was based on a cohort of elective surgeries for endometrioma performed by the same gynecologic team at Peking Union Medical College Hospital from January 2017 to October 2022. Patients diagnosed with previous spontaneously ruptured endometrioma during elective surgery were enrolled in the ruptured group. In the same cohort, patients with unruptured endometrioma treated during the same period were selected as the unruptured group by 1:2 matching according to age. Demographic and clinical information were collected and compared between two groups.

RESULTS

A total of 422 patients in the cohort were diagnosed with endometrioma. There were 38 patients (9.0%) in ruptured group and 76 patients in unruptured group. All enrolled participants were treated by laparoscopic surgery. In ruptured group, 86.8% patients had a history of acute abdominal pain, which was only 13.2% in unruptured group (P < 0.001). Compared to unruptured group, patients diagnosed with ruptured endometrioma had a lower BMI (P = 0.021), larger maximum diameter of endometrioma (P = 0.040), higher proportion of cul-de-sac partial obliteration rather than complete obliteration (P = 0.003).

CONCLUSIONS

Spontaneous rupture of endometrioma is not rare. The proportion of spontaneous rupture of endometrioma in our study was higher than that reported in the literatures. In women with endometrioma, the onset of acute abdominal pain should be considered a rupture of cyst, especially in patients with big cysts.

Ultrasound Features and Clinical Outcome of Patients with Ovarian Masses Diagnosed during Pregnancy: Experience of Single Gynecological Ultrasound Center

(1) Background: The number of adnexal masses detected during pregnancy has increased due to the use of first-trimester screening and increasingly advanced maternal age. Despite their low risk of malignancy, other risks associated with these masses include torsion, rupture and labor obstruction. Correct diagnosis and management are needed to guarantee both maternal and fetal safety. Adnexal masses may be troublesome to classify during pregnancy due to the increased volume of the uterus and pregnancy-related hormonal changes. Management should be based on ultrasound examination to provide the best treatment. The aim of this study was to describe the ultrasound features of ovarian masses detected during pregnancy and to optimize and personalize their management with the expertise of gynecologists, oncologists and sonographers. (2) Methods: Clinical, ultrasound, histological parameters and type of management (surveillance vs. surgery) were retrospectively retrieved. Patient management, perinatal outcomes and follow-up were also evaluated. (3) Results: according to the literature, these masses are most frequently benign, ultrasound follow-up is the best management, and obstetric outcomes are not considerably influenced by the presence of adnexal masses. (4) Conclusions: the management of patients with ovarian masses detected during pregnancy should be based on ultrasound examination, and a centralization in referral centers for ovarian masses should be considered.

Simple rules, O-RADS, ADNEX and SRR model: Single oncologic center validation of diagnostic predictive models alone and combined (two-step strategy) to estimate the risk of malignancy in adnexal masses and ovarian tumors

OBJECTIVE

To compare performance of Assessment of Different NEoplasias in the adneXa (ADNEX model), Ovarian-Adnexal Reporting and Data System (O-RADS), Simple Rules Risk (SRR) assessment and the two-step strategy based on the application of Simple Rules (SR) followed by SRR and SR followed by ADNEX in the pre-operative discrimination between benign and malignant adnexal masses (AMs).

METHODS

We conducted a retrospective study from January-2018 to December-2021 in which consecutive patients with at AMs were recruited. Accuracy metrics included sensitivity (SE) and specificity (SP) with their 95% confidence intervals (CI) were calculated for ADNEX, O-RADS and SRR. When SR was inconclusive a "two-step strategy" was adopted applying SR + ADNEX model and SR + SRR assessment.

RESULTS

A total of 514 women were included, 400 (77.8%) had a benign ovarian tumor and 114 (22.2%) had a malignant tumor. At a threshold malignancy risk of >10%, the SE and SP of ADNEX model, O-RADS and SRR were: 0.92 (95% CI, 0.86-0.96) and 0.88 (95% CI, 0.85-0.91); 0.93 (95% CI, 0.87-0.97) and 0.89 (95% CI, 0.96-0.92); 0.88 (95% CI, 0.80-0.93) and 0.84 (95% CI, 0.80-0.87), respectively. When we applied SR, 109 (21.2%) cases resulted inconclusive. The SE and SP of two-step strategy SR + SRR assessment and SR + ADNEX model were 0.88 (95% CI, 0.80-0.93) and 0.92 (95% CI, 0.89-0.94), SR + ADNEX model 0.90 (95% CI, 0.83-0.95) and 0.93 (95% CI, 0.90-0.96), respectively.

CONCLUSIONS

O-RADS presented the highest SE, similar to ADNEX model and SR + ADNEX model. However, the SR + ADNEX model presented the higher performance accuracy with the higher SP and PPV. This two-step strategy, SR and ADNEX model applicated to inconclusive SR, is convenient for clinical evaluation.

O-RADS US: A Systematic Review and Meta-Analysis of Category-specific Malignancy Rates

Background Ovarian-Adnexal Reporting and Data System (O-RADS) US provides a standardized method with which to stratify lesions into risk of malignancy categories, which is crucial for proper management. Purpose To perform a systematic review and meta-analysis to estimate malignancy rates for each O-RADS US score and evaluate the diagnostic performance of combined O-RADS US scores 4 and 5 in the diagnosis of malignancy. Materials and Methods A systematic literature search from the inception of the MEDLINE, EMBASE, and Web of Science databases through January 27, 2023, was performed for articles that reported using the O-RADS US stratification system and included malignancy rates per each O-RADS score. Bivariate random-effects models were used to determine the pooled malignancy rates for each O-RADS US score and to obtain summary estimates of the diagnostic performance of combined O-RADS US scores 4 and 5 in the diagnosis of malignant lesions. Results The final analysis included 18 studies consisting of 11 605 patients and 11 818 ovarian-adnexal lesions, with 2996 malignant (25.4%) and 8822 benign (74.6%) lesions. No malignant lesions were reported in O-RADS 1 category. The pooled percentages of malignancy were 0.6% (95% CI: 0.3, 1.0) for O-RADS 2, 3.9% (95% CI: 2.5, 5.4) for O-RADS 3, 43.5% (95% CI: 33.8, 53.2) for O-RADS 4, and 87.3% (95% CI: 83.0, 91.7) for O-RADS 5. The pooled sensitivity and specificity of combined O-RADS scores 4 and 5 in the diagnosis of malignant lesions were 95.6% (95% CI: 94.0, 97.2) and 76.6% (95% CI: 70.4, 82.7), respectively. Conclusion Each O-RADS US score provided the intended probability of malignant lesions as outlined by the O-RADS risk stratification system. When O-RADS US scores 4 and 5 were combined as a predictor for malignancy, O-RADS US showed a high sensitivity and moderate specificity. Clinical trial registration no. CRD42022352166 © RSNA, 2023 Supplemental material is available for this article.