Grading VUR with deep learning
| Article DOI | https://doi.org/10.1007/s10278-025-01438-1 |
| Objective | To grade VUR from VCUG images |
| AI Approach | CNN, independently analyzing the left and right urinary tracts |
| Data Source(s) | Single institutional series (1529 patients) |
| Model Input | VCUG images (768 x 768 pixels) |
| Model Outcome | VUR grade 0 vs 1-2 vs 3-5 |
| Model Metrics | AUC of 0.82; patient-level accuracy of 84% |
| Model Usability | Code and dataset not publicly available. Model structure detailed in journal article. |
AI = Artificial intelligence, VUR = Vesicoureteral reflux, AUC = Area under the receiver operator characteristic
AIPEDURO 
Leave a comment