A. Pertusa

Multimodal learning for imbalanced multi-label chest X-ray classification with radiology reports

This work investigates image-only, text-only, and multimodal learning on a curated subset of PadChest-GR, and demonstrates that simple concatenation-based fusion, when paired with appropriate balancing strategies and domain-specific encoders, provides a robust approach for imbalanced multi-label chest X-ray classificat...