Machine Learning Algorithms Incorporating Radiomic Ultrasound Features

NCT07793201 | Not Yet Recruiting

• 2 other identifiers: 27546RF-2024-12379537
• Study Type: observational
• Target: 12,000
• Locations: 1 country
• Sites: 1

Brief Summary

Adnexal masses represent a frequent clinical finding and their preoperative characterization remains challenging. Accurate discrimination between benign and malignant adnexal masses is essential to optimize patient management, avoid unnecessary surgery, and ensure appropriate referral of patients with suspected malignancy to specialized centers. This multicenter international observational study aims to evaluate the diagnostic performance and clinical utility of ultrasound-based machine learning (ML) models incorporating radiomic features for the characterization of adnexal masses. The study will develop and validate artificial intelligence (AI)-based models using ultrasound imaging data to support the preoperative classification of adnexal masses. The primary objective of the study is to evaluate the ability of ultrasound-based ML models to distinguish between benign and malignant adnexal masses. Secondary objectives include the evaluation of additional AI-based classification models among masses identified as malignant, including the discrimination between borderline tumors, primary invasive malignancies, and metastatic lesions. Furthermore, the study will assess the ability of AI models to differentiate primary epithelial ovarian carcinoma from non-epithelial ovarian malignancies among cases classified as primary ovarian cancer. The clinical utility of the developed models will be assessed using decision curve analysis. In addition, a retrospective post hoc evaluation will be performed in an independent prospective external validation cohort to explore the potential clinical impact of an AI-based preoperative model for the management of adnexal masses. This evaluation will compare AI model outputs with actual clinical decisions made during routine care, without influencing patient management or altering the diagnostic and therapeutic pathway. The post hoc clinical impact analysis will assess diagnostic concordance between AI predictions and clinicians' preoperative assessments, the potential proportion of avoidable surgical procedures according to AI model predictions, surgical and follow-up complications, and cost-effectiveness through comparison of healthcare resource utilization between standard clinical management and a reconstructed AI-supported scenario. Patient-reported outcomes will also be evaluated, including patient satisfaction regarding diagnostic communication, clarity of information provided, and perceived quality of care within the standard clinical management pathway. Overall, this study aims to investigate whether ultrasound-based AI models integrating radiomic features can improve the characterization of adnexal masses and provide clinically useful tools to support personalized and efficient patient management.

Conditions

Adnexal Masses

Keywords

Adnexal masses; Ovarian cancer; Ultrasound; Radiomics; Ultrasonography; Artificial intelligence

Outcome Measures

Primary Outcomes (1)

• Diagnostic performance of an ultrasound-based machine learning model for discrimination of benign and malignant adnexal masses

  Diagnostic performance of a machine learning model incorporating ultrasound radiomic features for distinguishing benign from malignant adnexal masses, assessed against the reference diagnosis. Performance will be evaluated using area under the receiver operating characteristic curve (AUC-ROC), accuracy, sensitivity, specificity, and calibration.

  At final diagnosis or completion of 1-year follow-up

Study Arms (2)

Development Cohort

Patients with adnexal masses evaluated at participating centers with available ultrasound imaging data and reference diagnosis. Data from this cohort will be used for the development and internal validation of ultrasound-based machine learning models incorporating radiomic features.

External Validation Cohort

Independent prospective cohort of patients with adnexal masses enrolled at participating centers. This cohort will be used for external validation of the developed ultrasound-based machine learning models and assessment of their performance and generalizability. AI model outputs will not influence clinical management.

Eligibility Criteria

• Age: 18 Years+
• Healthy Volunteers: No
• Age Groups: Adult (18-64), Older Adult (65+)
• Sampling Method: Non-Probability Sample

Study Population

Patients with adnexal masses

You may qualify if:

• Patients with an adnexal mass identified at ultrasound examination who either undergo surgery within 6 months of the ultrasound or have at least 1 year of follow-up. This includes: the retrospective cohort recruited from IOTA centers, and the prospective cohort recruited from nonIOTA centers.
• Age ≥ 18 years old
• Availability of at least one grayscale digital ultrasound image clearly depicting the adnexal mass.
• Signed written informed consent

You may not qualify if:

• Patients without available digital ultrasound images.
• Patients without an outcome (final histology or follow up at one year).
• Absence of written informed consent

Sponsors & Collaborators

• Fondazione Policlinico Universitario Agostino Gemelli IRCCS: lead

Study Sites (1)

Fondazione Policlinico Universitario Agostino Gemelli IRCCS

Roma, 00168, Italy

Location

MeSH Terms

Conditions

Ovarian Neoplasms

Condition Hierarchy (Ancestors)

Endocrine Gland Neoplasms; Neoplasms by Site; Neoplasms; Ovarian Diseases; Adnexal Diseases; Genital Diseases, Female; Female Urogenital Diseases; Female Urogenital Diseases and Pregnancy Complications; Urogenital Diseases; Genital Neoplasms, Female; Urogenital Neoplasms; Genital Diseases; Endocrine System Diseases; Gonadal Disorders

Study Officials

• Antonia Carla Testa

  Fondazione Policlinico Universitario Agostino Gemelli IRCCS

  PRINCIPAL INVESTIGATOR

Central Study Contacts

Antonia Carla Testa, Professor

CONTACT

+390630156399; antoniacarla.testa@policlinicogemelli.it

Study Design

• Study Type: observational
• Observational Model: COHORT
• Time Perspective: CROSS SECTIONAL
• Target Duration: 12 Months
• Sponsor Type: OTHER
• Responsible Party: SPONSOR

Study Record Dates

• First Submitted: July 31, 2026
• First Posted: August 28, 2026
• Study Start (Estimated): September 1, 2026
• Primary Completion (Estimated): January 1, 2028
• Study Completion (Estimated): September 1, 2029
• Last Updated: August 28, 2026

Record last verified: 2026-07

Data Sharing

• IPD Sharing: Will not share

Locations

IT (1)