C67 C66 C65
Conditions
Interventions
Group 1: AI training and evaluation using archived urinary cytology slides from patients with positive cytopathological findings (presence of high-grade urothelial cells, suspicious for high-grade uro
Sponsors
Klinik für Urologie und Urochirurgie der Universitätsmedizin Mannheim
Eligibility
Sex/Gender
All
Age
18 Years to No maximum
Inclusion criteria
Inclusion criteria: -Archived urinary cytology slides from routine clinical diagnostics with documented cytopathological findings (positive, i.e. high-grade / suspicious for high-grade urothelial carcinoma [SHGUC], or negative). -For cases with ambiguous cytopathological findings, in particular the presence of low-grade urothelial cells, availability of additional histopathological findings or a documented endoscopic exclusion of urothelial carcinoma (e.g., cystoscopy, URS, TUR-P, DJ catheter placement/exchange, urethrotomy) within close temporal proximity to urinary cytology.
Exclusion criteria
Exclusion criteria: -Urinary cytology samples obtained from neobladders. -Slides categorized as “unsatisfactory for evaluation” according to the Paris System for Reporting Urinary Cytology. -Cases with ambiguous cytopathological findings (i.e., presence of low-grade urothelial cells) without available histopathological or endoscopic reference findings. -Insufficient image quality after digitization (e.g., blurred or non-focusable scans).
Design outcomes
Primary
| Measure | Time frame |
|---|---|
| What: Diagnostic performance of the AI model for the detection of urothelial carcinoma based on urinary cytology slides. When: Assessed during model training and validation, including internal validation using a randomly split validation dataset (approximately 25% of the retrospective dataset) and external prospective validation on an independent dataset collected between April 2025 and January 2026. How: Evaluation of AI-based slide-level predictions against the reference standard (ground truth) defined by cytopathological findings in the retrospective dataset and by histological or endoscopic findings in the prospective validation dataset, with performance quantified by the area under the receiver operating characteristic curve (AUC-ROC). | — |
Secondary
| Measure | Time frame |
|---|---|
| Sensitivity and specificity of the AI model for the detection of urothelial carcinoma based on urinary cytology slides, assessed separately in the internal validation and in the external prospective validation. | — |
Countries
Germany
Contacts
Public ContactGloria Baumann
Klinik für Urologie und Urochirurgie der Universitätsmedizin Mannheim
Outcome results
None listed