Abstract
Medical image segmentation models are often trained on curated datasets, leading to performance degradation when deployed in real-world clinical settings due to mismatches between training and test distributions. While data augmentation techniques are widely used to address these challenges, traditional visually consistent augmentation strategies lack the robustness needed for diverse real-world scenarios. In this work, we systematically evaluate alternative augmentation strategies, focusing on MixUp and Auxiliary Fourier Augmentation. These methods mitigate the effects of multiple variations without explicitly targeting specific sources of distribution shifts. We demonstrate how these techniques significantly improve out-of-distribution generalization and robustness to imaging variations across a wide range of transformations in cardiac cine MRI and prostate MRI segmentation. We quantitatively find that these augmentation methods enhance learned feature representations by promoting separability and compactness. Additionally, we highlight how their integration into nnU-Net training pipelines provides an easy-to-implement, effective solution for enhancing the reliability of medical segmentation models in real-world applications.
Original language | English |
---|---|
Publisher | ArXiv.org |
DOIs | |
Publication status | Published - 15 May 2025 |
Keywords
- cs.CV
- cs.LG
Fingerprint
Dive into the research topics of 'Data-Agnostic Augmentations for Unknown Variations: Out-of-Distribution Generalisation in MRI Segmentation'. Together they form a unique fingerprint.Datasets
-
Data-Agnostic Augmentations for Unknown Variations: Out-of-Distribution Generalisation in MRI Segmentation - Pretrained Models and Test Data
Vaish, P. (Creator), Meister, F. (Creator), Heimann, T. (Creator), Brune, C. (Creator) & Wolterink, J. (Creator), Zenodo, 15 May 2025
DOI: 10.5281/zenodo.15517158, https://zenodo.org/records/15517158 and 3 more links, https://doi.org/10.5281/zenodo.15517159, https://zenodo.org/records/15517159, https://arxiv.org/abs/arXiv:2505.10223 (show fewer)
Dataset