A domain-adaptive deep contrastive network for magnetic resonance imaging-driven bladder cancer classification - PubMed
6 hours ago
- #Medical Imaging
- #Deep Learning
- #Bladder Cancer
- Bladder cancer is a prevalent malignancy with high morbidity and mortality.
- Deep learning shows promise for automated bladder cancer classification using MRI.
- Challenges include inter-center distribution discrepancies and feature discriminability between NMIBC and MIBC.
- Proposed solution: Domain-Adaptive Deep Contrastive Network (DADCNet) for improved classification.
- DADCNet enhances cross-center generalization and inter-class separability.
- Achieves high accuracy (0.955), F1-score (0.955), and AUC (0.991) on multi-center datasets.