CVPR Findings2026
MAE-XNT: A Foundation Model for Segmenting Neuronal Tissue Volumes Generated with X-Ray Nanotomography
Alfred Laugros, Sebastien Roig, Alexandra Pacureanu
Abstract
Mapping neural circuits in brain tissue volumes provides essential information to better understand how the brain works. X-ray nanotomography (XNT) is emerging as a new technology for scalable imaging of neuronal tissue volumes at the nanoscale. Wiring diagrams can be reconstructed from the acquired image volumes using deep learning models trained on data annotated by experts. However, producing enough annotations for training a deep learning model to perform well across extensive brain tissue volumes is extremely costly. In this paper, we use self-supervised learning to produce the first foundation model for segmenting brain tissue volumes acquired with XNT. We call the model MAE-XNT, and we demonstrate that it drastically reduces the amount of expert annotation required to segment a new volume. We also show that our foundation model can be used to produce more accurate segmentation maps under distribution shifts. By fine-tuning MAE-XNT, we obtain the best segmentation performance on an axon dataset and a cell nuclei dataset.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- Segment Anything cites An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
- Segment Anything cites Learning Transferable Visual Models From Natural Language Supervision
- An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale cites A Simple Framework for Contrastive Learning of Visual Representations
- Bootstrap Your Own Latent - A New Approach to Self-Supervised Learning cites A Simple Framework for Contrastive Learning of Visual Representations
- Learning Transferable Visual Models From Natural Language Supervision cites A Simple Framework for Contrastive Learning of Visual Representations
- Learning Transferable Visual Models From Natural Language Supervision cites An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
- Learning Transferable Visual Models From Natural Language Supervision cites Bootstrap Your Own Latent - A New Approach to Self-Supervised Learning
Builds on13
- Learning Transferable Visual Models From Natural Language SupervisionAlec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh et al.ICML 2021 · 47,906 citations
- A Simple Framework for Contrastive Learning of Visual RepresentationsTing Chen, Simon Kornblith, Mohammad Norouzi, Geoffrey E. HintonICML 2020 · 24,064 citations
- An Image is Worth 16x16 Words: Transformers for Image Recognition at ScaleAlexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn et al.ICLR 2021 · 21,477 citations
- Segment AnythingAlexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao et al.ICCV 2023 · 13,211 citations
- Bootstrap Your Own Latent - A New Approach to Self-Supervised LearningJean-Bastien Grill, Florian Strub, Florent Altché, Corentin Tallec et al.NeurIPS 2020 · 9,171 citations
Related papers
- VasoMIM: Vascular Anatomy-Aware Masked Image Modeling for Vessel SegmentationDe-Xing Huang, Xiao-Hu Zhou, Mei-Jiang Gui, Xiao-Liang Xie et al.AAAI 2026 · 2 citations
- Revisiting MAE Pre-training for 3D Medical Image SegmentationTassilo Wald, Constantin Ulrich, Stanislav Lukyanenko, Andrei Goncharov et al.CVPR 2025
- SynAnno: Interactive Guided Proofreading of Synaptic AnnotationsLeander Lauenburg, Jakob Troidl, Adam Gohain, Zudi Lin et al.IEEE VIS 2025
- Learning Generalizable 3D Medical Image Representations from Mask-Guided Self-SupervisionYunhe Gao, Yabin Zhang, Chong Wang, Jiaming Liu et al.CVPR 2026
- An OpenMind for 3D Medical Vision Self-supervised LearningTassilo Wald, Constantin Ulrich, Jonathan Suprijadi, Sebastian Ziegler et al.ICCV 2025 · 5 citations
