Paper accepted at MICCAI 2025

Our paper PSAT: Pediatric Segmentation Approaches via Adult Augmentations and Transfer Learning has been accepted at MICCAI 2025 in Daejeon, South Korea.

In this work, we investigate how different dataset fingerprints, learning sets, data augmentation, and transfer learning strategies (fine-tuning vs. continual learning) affect segmentation performance on pediatric CT scans. We show that continual learning strategies improve generalization across diverse pediatric datasets.

• Preprint: arXiv:2507.05764
• Code: github.com/ICANS-Strasbourg/PSAT
• Models: Hugging Face