Foundation Model-based AI for Preoperative Three-class Ovarian Tumour Classification: Using Computer Aided Diagnosis (CADx) to Predict Whether an Ovarian Tumour is Benign, Malignant or Borderline
| Registry | NCT07883551, acronym OvaClass |
|---|---|
| Status | Not yet recruiting |
| Study type | Observational |
| Enrollment | 1300 (Estimated) |
| Sponsor | Catharina Ziekenhuis Eindhoven (Other); collaborators: Eindhoven University of Technology, Maastricht University, Maastricht University Medical Center, Amphia Hospital, Leiden University Medical Center, Amsterdam UMC, location VUmc, University Hospital, Ghent, Antoni van Leeuwenhoek Hospital |
| Conditions | Ovarian Cancer, Ovarian Tumor, Borderline Ovarian Tumor |
| Dates | start 2026-11-01 (estimated), primary completion 2031-10-01 (estimated), completion 2031-10-01 (estimated), first posted 2026-10-20, last updated 2026-10-20 |
| Locations | 7 sites in Belgium, Netherlands |
Related Papers Related Patents Related Grants Related Experts
All trials for Ovarian Cancer All trials for Ovarian Tumor
The goal of this observational study is to learn if a computer program can tell what type of ovarian tumour a person has from a CT scan taken before surgery. The program uses artificial intelligence (AI). It sorts tumours into three groups:
* benign: not cancer
* borderline: abnormal cells that usually behave less aggressively than cancer
* malignant: cancer
Before surgery, doctors try to judge which group a tumour belongs to. This guides where and how the operation is done. This judgement is often hard, especially for borderline tumours. An AI program that does it well could help doctors plan surgery.
The main questions it aims to answer are:
* How well does the AI program sort ovarian tumours into benign, borderline and malignant on CT scans?
* Does the AI program work as well on scans from hospitals it did not learn from?
Researchers will compare the AI program's results with the final diagnosis from the tumour tissue removed during surgery. They will also compare the program with methods doctors use today, such as ultrasound-based risk scores and the radiologist's report. In a smaller group, researchers will test the program on MRI scans.
This study only uses information that already exists. It includes adults who had surgery for a mass in or near the ovary at hospitals in the Netherlands and Belgium. Researchers will use:
* CT or MRI scans made before surgery as part of regular care
* information from medical records
* the tissue diagnosis after surgery
Participants will not be contacted. They will not have any extra scans, tests or visits. Each hospital codes the data before sharing them, so researchers cannot see participants' names. The study does not change anyone's care.
* Background and rationale Preoperative distinction between benign, borderline and malignant ovarian tumours determines referral, surgical approach and extent of surgery. Ultrasound-based models such as IOTA-ADNEX and the Risk of Malignancy Index (RMI) are the current triage tools, but their performance varies between populations and users. CT is often performed in indeterminate cases or for staging. There are no validated diagnostic criteria for CT, and interpretation depends on the reader. Published CT-based computer-aided diagnosis (CADx) models are mostly single-centre, binary (benign versus malignant), dependent on manual tumour segmentation and rarely externally validated. Borderline tumours need different management but are usually excluded or merged with malignant tumours. Earlier work from the Catharina Cancer Institute and Eindhoven University of Technology (TU/e) showed considerable domain shift between hospitals in a multicentre CT dataset (Koch et al., 2026).
* Index test The index test is a CADx pipeline built on SPECTRE, a CT foundation model developed at TU/e and pre-trained with self-supervised learning on more than 250,000 CT scans. SPECTRE is fine-tuned with parameter-efficient adaptation. The pre-trained encoder is kept frozen or partially unfrozen, and lightweight task heads are trained for tumour localisation and three-class classification. The pipeline works directly on the CT volume. Tumour localisation is learned within the model, so no tumour masks or bounding boxes are needed for training or evaluation. An uncertainty quantification module produces calibrated confidence estimates alongside each prediction. The overall architecture is fixed. Training details such as the fine-tuning schedule, loss weighting and calibration method may be adapted based on interim results. Radiomics-based machine learning (handcrafted features with classifiers such as logistic regression, support vector machines, random forests or gradient boosting) is used only as a benchmark comparator. The model is applied retrospectively and its output is never available to treating clinicians.
* Reference standard Definitive postoperative histopathology, classified as benign, borderline or malignant according to the 2020 WHO Classification of Female Genital Tumours. Ambiguous cases are adjudicated by an experienced gynaecological pathologist to keep labelling consistent across centres. In bilateral tumours with discordant histology, a predefined rule is applied (for example, selection of the lesion with the highest malignant potential).
* Data acquisition and preprocessing Participating centres identify eligible patients and pseudonymise and code imaging, clinical and pathology data locally before transfer under formal data-sharing agreements. The key linking codes to identities stays at the local centre. Images are exported in DICOM format and may be converted to volumetric formats (NIfTI or NRRD). CT preprocessing includes intensity windowing (for example -100 to +300 HU), normalisation and voxel resampling to a standard resolution. Scanner manufacturer and model, slice thickness, contrast phase and reconstruction kernel are recorded for analysis of domain shift. For the MRI pilot, T2-weighted, T1-weighted with and without fat suppression, diffusion-weighted with ADC maps and contrast-enhanced T1-weighted sequences are used when available.
* Model development and validation Within the development cohort, stratified cross-validation (or an equivalent resampling method adapted to the final sample size) is used, stratified by outcome class. Hyperparameter tuning and model selection take place only within the training data of each resampling iteration. All scans and lesions from one patient are kept in the same split to prevent data leakage. External validation uses one or more centres that are excluded entirely from model development. The allocation of centres to development and external validation is documented before any performance analysis. Calibration methods such as Platt scaling or temperature scaling may be applied.
* Statistical analysis Discrimination is assessed with one-vs-rest AUROC per class and macro- and micro-averaged multiclass AUROC. Classification metrics include sensitivity, specificity, positive and negative predictive values, F1-score, balanced accuracy and confusion matrices. 95% confidence intervals are estimated by bootstrapping where appropriate. Calibration is assessed with calibration curves and the Brier score. Net benefit across clinically relevant risk thresholds is assessed with decision curve analysis. Comparison with RMI, IOTA-ADNEX, radiologist reports and, in the MRI pilot, O-RADS MRI is restricted to patients in whom the comparator is available or can be reconstructed. Paired comparisons use the DeLong test for ROC curves, McNemar's test for paired proportions, kappa statistics for agreement and the Brier score for probabilistic predictions. Baseline characteristics are compared with chi-squared or Fisher's exact tests and t-tests or Mann-Whitney U tests as appropriate. Two-sided p-values are reported with a nominal significance level of 0.05, with emphasis on effect sizes and confidence intervals. Missing clinical data are described and not imputed. Model behaviour is explored with attention and saliency maps (for example Grad-CAM).
* Exploratory analyses Where data can be linked in coded form and are covered by ethics approval and data-sharing agreements, exploratory analyses will assess whether adding blood-based biomarkers (CA-125, CA-125 glycovariants, CEA, ctDNA, proteomics) or radiologist report content to CT imaging changes classification performance. In a subset with digitised histopathology slides, AI-based analysis of whole slide images will be explored to find histological correlates of CT-based tumour characteristics. These analyses do not change the primary imaging-based objective.
* Reporting Results will be reported following TRIPOD+AI, supplemented by CLAIM and STROBE where applicable.
| Arm | Type | Interventions | Description |
|---|---|---|---|
| CT development cohort | * Description: Adult patients with an adnexal mass who had preoperative CT and surgery with definitive histopathology at the development centres (Catharina Hospital Eindhoven, Amphia Hospital, Netherlands Cancer Institute-Antoni van Leeuwenhoek, Leiden University Medical Centre and Amsterdam UMC). Includes patients from a previous multicentre CADx cohort and the OVIDETECT registry, approximately 930 scans, of which around 118 are borderline tumours. Used for model training, tuning and internal cross-validation. * Intervention: SPECTRE-based CADx model | ||
| CT external validation cohort | * Description: Adult patients meeting the same criteria at centres held out entirely from model development (anticipated: Maastricht UMC+ and/or Ghent University Hospital), approximately 400 scans. Used only to assess generalisability across institutions. Centre allocation is fixed and documented before analysis. * Intervention: SPECTRE-based CADx model | ||
| MRI pilot cohort (exploratory) | * Description: Adult patients from participating centres with a preoperative MRI of the adnexal mass and definitive histopathology, approximately 100 to 200 scans. Used to explore preliminary CADx feasibility on MRI and to compare with O-RADS MRI. Not part of the primary CT-based objective. * Intervention: SPECTRE-based CADx model |
Sex: Female, ages: 18 Years to any
Inclusion Criteria: * Age 18 years or older at the time of imaging * Ovarian or adnexal mass on clinical and/or radiological assessment that can be reliably identified as the index lesion on CT and/or MRI * Preoperative CT and/or MRI scan available, acquired as part of routine clinical care * Surgical removal of the ovarian or adnexal mass, with definitive postoperative histopathology available as reference standard * Bilateral ovarian tumours are eligible. If histopathology differs between sides, a predefined rule is applied (for example, selection of the lesion with the highest malignant potential) * Advanced ovarian malignancy, including ascites or peritoneal disease, is eligible if the primary adnexal mass can be clearly identified on imaging Exclusion Criteria: * No definitive postoperative histopathology available * Imaging of insufficient diagnostic quality (for example incomplete series, corrupted files or non-diagnostic sequences) * Extra-ovarian disease that prevents reliable identification or attribution of the adnexal lesion as the index tumour (for example diffuse peritoneal carcinomatosis or a confirmed non-ovarian primary malignancy) * Previous ovarian malignancy or prior oncological treatment affecting the adnexal region * Pathology report judged indefinite or inconclusive by the reviewing clinician
| Facility | City | Country | Status |
|---|---|---|---|
| Ghent University Hospital | Ghent | Belgium | |
| Netherlands Cancer Institute - Antoni van Leeuwenhoek | Amsterdam | Netherlands | |
| Amsterdam University Medical Centre | Amsterdam | Netherlands | |
| Amphia Ziekenhuis | Breda | Netherlands | |
| Catharina Ziekenhuis | Eindhoven | Netherlands | |
| Leiden University Medical Centre | Leiden | Netherlands | |
| Maastricht UMC+ | Maastricht | Netherlands |