ESTRO 2025 - Abstract Book

S2534

Physics - Autosegmentation

ESTRO 2025

Conclusion: The nnU-Net model outperformed the SynthSeg on the internal dataset in both segmentation metrics and clinician ratings. While segmentation metrics showed no significant differences between the models on the external set, clinician ratings favored nnU-Net, suggesting enhanced clinical acceptability. These findings indicate that while custom model training can provide advantages, pre-trained models like SynthSeg may be sufficient for most clinical applications.

Keywords: nnU-net, Ventricle and periventricular space