A single biomedical image can be meaningfully segmented in multiple ways, depending on the desired application. For instance, a brain MRI can be segmented according to tissue types, vascular territories, broad anatomical regions, fine-grained anatomy, or pathology, etc. Existing automatic segmentation models typically either (1) support only a single protocol, the one they were trained on, or (2) require labor-intensive manual prompting to specify the desired segmentation. We introduce Pancakes, a framework that, given a new image from a previously unseen domain, automatically generates multi-label segmentation maps for multiple plausible protocols, while maintaining semantic consistency across related images. Pancakes introduces a new problem formulation that is not currently attainable by existing foundation models. In a series of experiments on seven held-out datasets, we demonstrate that our model can significantly outperform existing foundation models in producing several plausible whole-image segmentations, that are semantically coherent across images.
@article{arxiv.2512.13534,
title = {Pancakes: Consistent Multi-Protocol Image Segmentation Across Biomedical Domains},
author = {Marianne Rakic and Siyu Gai and Etienne Chollet and John V. Guttag and Adrian V. Dalca},
journal= {arXiv preprint arXiv:2512.13534},
year = {2025}
}
Comments
Accepted at NeurIPS 2025. Code available at: https://github.com/mariannerakic/Pancakes