Abstract:Objective: Our primary aim was to identify radiomic ultrasound features that can distinguish benign from malignant adnexal masses with solid ultrasound morphology, and primary invasive from 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. Our secondary aim was to compare the diagnostic performance of our radiomics models with that of the ADNEX model and subjective assessment by an experienced ultrasound examiner. Methods: This is a retrospective observational single center study. Patients with a histological diagnosis of an adnexal tumor with solid morphology at preoperative ultrasound examination performed between 2014 and 2021 were included. The patient cohort was split into training and validation sets with a ratio of 70:30 and with the same proportion of benign and malignant (borderline, primary invasive and metastatic) tumors in the two subsets. The extracted radiomic features belonged 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's age, and those radiomic features that were statistically significantly different between benign and malignant adnexal masses (Wilcoxon-Mann-Whitney Test with Benjamini-Hochberg correction for multiple comparisons) and assessed as not redundant based on the Pearson correlation coefficient. We describe discriminative ability as area under the receiver operating characteristics curve (AUC) and classification performance as sensitivity and specificity. Results: 326 patients were identified and 775 preoperative ultrasound images were analyzed. 68 radiomic features were extracted, 52 differed statistically significantly between benign and malignant tumors in the training set, and 18 features were selected for inclusion in model building. The same 52 radiomic features differed statistically significantly between benign, primary invasive malignant and metastatic tumors. However, the values of the features manifested overlap between primary malignant and metastatic tumors and did not differ statistically significantly between them. In the validation set, 25/98 tumors (25.5%) were benign, 73/98 (74.5%) were malignant (6 borderline, 57 primary invasive, 10 metastases). In the validation set, a model including only radiomics features had an AUC of 0.80, and 78% sensitivity and 76% specificity at its optimal risk of malignancy cutoff (68% based on Youden's index). The corresponding results for a model including age and radiomics features were 0.79, 86% and 56% (cutoff 60% based on Youden's method), while those of the ADNEX model were 0.88, 99% and 64% (at 20% malignancy cutoff). Subjective assessment had sensitivity 99% and specificity 72%. Conclusions: Even though our radiomics models had discriminative ability inferior to that of the ADNEX model, our results are promising enough to justify continued development of radiomics analysis of ultrasound images of adnexal masses. This article is protected by copyright. All rights reserved.

Evaluation of adnexal masses with three-dimensional ultrasonography

Three-dimensional sonographic morphologic assessment in complex adnexal masses: preliminary experience

Imaging Evaluation of Ovarian Masses

Evaluation of Primary Adnexal Masses by 3T MRI: Categorization with Conventional MR Imaging and Diffusion-Weighted Imaging.

Application of Ultrasound Scores (Subjective Assessment, Simple Rules Risk Assessment, ADNEX Model, O-RADS) to Adnexal Masses of Difficult Classification

Imaging techniques for the pre-surgical diagnosis of adnexal tumours

Diagnostic accuracy of transvaginal ultrasound examination for assigning a specific diagnosis to adnexal masses

Prospective Evaluation of 3T MRI Findings for Primary Adnexal Lesions and Comparison with the Final Histological Diagnosis.

Role of 3-Dimensional Sonography in the Assessment of Submucous Fibroids: A Pilot Study

A new computer-aided diagnostic tool for non-invasive characterisation of malignant ovarian masses: results of a multicentre validation study

Imaging Strategy for Early Ovarian Cancer: Characterization of Adnexal Masses with Conventional and Advanced Imaging Techniques

Amide Proton Transfer Weighted MRI in Differential Diagnosis of Ovarian Masses with Cystic Components: A Preliminary Study.

Value of transvaginal three-dimensional power Doppler ultrasound parameters for discrimination between malignant and benign ovarian masses

Role of Three-Dimensional Transvaginal Sonography in Evaluation of the Depth of Myometrial Invasion of Endometrial Carcinoma

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

Three-dimensional Ultrasonography for the Prediction of Breast Cancer Prognosis.

Conventional, Doppler and Contrast-Enhanced Ultrasonography in Differential Diagnosis of Ovarian Masses

Evaluation of Selected MRI Parameters in the Differentiation of Mucinous Ovarian Carcinomas and Metastatic Ovarian Tumors

Evaluation of primary adnexal masses with 3.0T MRI:value of conventional MR imaging and diffusion-weighted imaging in categorization

Primary versus Secondary Ovarian Malignancy: Imaging Findings of Adnexal Masses in the Radiology Diagnostic Oncology Group Study

Contrast-enhanced sonography in the examination of benign and malignant adnexal masses