Urothelial carcinoma of the urinary bladder and upper urinary tract
Conditions
Interventions
Group 1: AI training and evaluation using prospectively collected urinary cytology slides from patients with histologically confirmed urothelial carcinoma of the urinary bladder or upper urinary tract
Sponsors
Universitätsmedizin Mannheim
Eligibility
Sex/Gender
All
Age
18 Years to No maximum
Inclusion criteria
Inclusion criteria: -age = 18 years and ability to provide informed consent -suspected urothelial carcinoma of the urinary bladder or upper urinary tract -histologically confirmed urothelial carcinoma of the urinary bladder or upper urinary tract -planned cystoscopic or endourological examination -ability to provide at least one urine sample
Exclusion criteria
Exclusion criteria: -age < 18 years -inability to provide informed consent -inability to obtain a urine sample
Design outcomes
Primary
| Measure | Time frame |
|---|---|
| What: Sensitivity of the AI model for the detection of urothelial carcinoma of the urinary bladder or upper urinary tract based on prospectively collected urinary cytology slides. When: Assessed during training and evaluation of the AI model after completion of sample collection, processing, and digitization. How: Comparison of the AI-based classification of urinary cytology slides with the reference standard (“ground truth”). A positive reference standard is defined as histologically confirmed urothelial carcinoma of the urinary bladder or upper urinary tract. | — |
Secondary
| Measure | Time frame |
|---|---|
| -Specificity of the AI model for the detection of urothelial carcinoma of the urinary bladder or upper urinary tract. A negative reference standard is defined by a negative histopathological result or documented endoscopic exclusion of urothelial carcinoma. -Diagnostic performance of the AI model, particularly based on the area under the receiver operating characteristic curve (AUC-ROC), positive and negative predictive values, and balanced accuracy. -Comparison of diagnostic performance across different staining techniques used for urinary cytology slides, particularly with regard to sensitivity, specificity, and AUC-ROC. -Comparison of single and serially collected urine samples with regard to sensitivity and specificity for the detection of urothelial carcinoma. -Identification and analysis of characteristic cytological features and image patterns associated with the presence of urothelial carcinoma. -Assessment of the biological and cytomorphological plausibility of AI-based decisions using the image regions with the highest attention scores, for example the top 50 attention patches. The analysis will assess whether diagnostically relevant cells or cellular patterns were identified and contributed to the AI-based classification. | — |
Countries
Germany
Contacts
Public ContactGloria Baumann
Universitätsmedizin Mannheim
Outcome results
None listed