mlsanityData to 13 Oct 2026

Learning to Upscale 3D Segmentations in Neuroimaging

arXiv 2505.21697 ↗First posted 27 May 2025cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Obtaining high-resolution (HR) segmentations from coarse annotations is a pervasive challenge in computer vision. Applications include inferring pixel-level segmentations from token-level labels in vision transformers, upsampling coarse masks to full resolution, and transferring annotations from legacy low-resolution (LR) datasets to modern HR imagery. These challenges are especially acute in 3D neuroimaging, where manual labeling is costly and resolutions continually increase. We propose a scalable framework that generalizes across resolutions and domains by regressing signed distance maps, enabling smooth, boundary-aware supervision. Crucially, our model predicts one class at a time, which substantially reduces memory usage during training and inference (critical for large 3D volumes) and naturally supports generalization to unseen classes. Generalization is further improved through training on synthetic, domain-randomized data. We validate our approach on ultra-high-resolution (UHR) human brain MRI (~100 μm), where most existing methods operate at 1 mm resolution. Our framework effectively upsamples such standard-resolution segmentations to UHR detail. Results on synthetic and real data demonstrate superior scalability and generalization compared to conventional segmentation methods. Code is available at: https://github.com/HuXiaoling/Learn2Upscale.

Key facts. No citing papers in the map yet (data through 13 Oct 2026). Builds on 43 of its 73 references that are in the map. First posted on arXiv 27 May 2025.

Builds on

43 of 73 references.

References · 43

  1. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,043
  2. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,656
  3. 2023Segment Anythingcited by 8,195
  4. 2014Fully Convolutional Networks for Semantic Segmentationcited by 4,849
  5. 2016DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFscited by 3,240
  6. 2019DeepSDF: Learning Continuous Signed Distance Functions for Shape Representationcited by 2,686
  7. 2016V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentationcited by 2,287
  8. 2017Rethinking Atrous Convolution for Semantic Image Segmentationcited by 2,159
  9. 20163D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotationcited by 1,475
  10. 2017Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real Worldcited by 1,411
  11. 2019Panoptic Feature Pyramid Networkscited by 644
  12. 2017Generalised Dice overlap as a deep learning loss function for highly unbalanced segmentationscited by 615
31 more
  1. 2014Semantic Image Segmentation with Deep Convolutional Nets and Fully Connected CRFscited by 553
  2. 2015Learning Deconvolution Network for Semantic Segmentationcited by 365
  3. 2020Neural Unsigned Distance Fields for Implicit Function Learningcited by 265
  4. 2016Efficient Multi-Scale 3D CNN with Fully Connected CRF for Accurate Brain Lesion Segmentationcited by 263
  5. 2021SynthSeg: Segmentation of brain MRI scans of any contrast and resolution without retrainingcited by 236
  6. 2020Modeling the Background for Incremental Learning in Semantic Segmentationcited by 189
  7. 2019Models Genesis: Generic Autodidactic Models for 3D Medical Image Analysiscited by 159
  8. 2019SDFDiff: Differentiable Rendering of Signed Distance Fields for 3D Shape Optimizationcited by 159
  9. 2015Brain Tumor Segmentation with Deep Neural Networkscited by 137
  10. 2020SynthMorph: learning contrast-invariant registration without acquired imagescited by 137
  11. 2020CascadePSP: Toward Class-Agnostic and Very High-Resolution Segmentation via Global and Local Refinementcited by 113
  12. 2019FastSurfer -- A fast and accurate deep learning based neuroimaging pipelinecited by 110
  13. 2017Transfer Learning for Domain Adaptation in MRI: Application in Brain Lesion Segmentationcited by 66
  14. 2021Recurrent Mask Refinement for Few-Shot Medical Image Segmentationcited by 63
  15. 2014Multi-Atlas Segmentation of Biomedical Images: A Surveycited by 56
  16. 2019Deep Distance Transform for Tubular Structure Segmentation in CT Scanscited by 52
  17. 2022Robust machine learning segmentation for large-scale analysis of heterogeneous clinical brain MRI datasetscited by 51
  18. 2021Incremental Few-Shot Instance Segmentationcited by 49
  19. 2023Towards Better Gradient Consistency for Neural Signed Distance Functions via Level Set Alignmentcited by 39
  20. 2021High Quality Segmentation for Ultra High-resolution Imagescited by 31
  21. 2022GenSDF: Two-Stage Learning of Generalizable Signed Distance Functionscited by 30
  22. 2023Unsupervised Inference of Signed Distance Functions from Single Sparse Point Clouds without Learning Priorscited by 28
  23. 2021Multi-Domain Incremental Learning for Semantic Segmentationcited by 25
  24. 2024FlowSDF: Flow Matching for Medical Image Segmentation Using Distance Transformscited by 25
  25. 2024CPM: Class-conditional Prompting Machine for Audio-visual Segmentationcited by 11
  26. 2023Brain-ID: Learning Contrast-agnostic Anatomical Representations for Brain Imagingcited by 10
  27. 2019Distance transform regression for spatially-aware deep semantic segmentationcited by 9
  28. 2021Inferring the Class Conditional Response Map for Weakly Supervised Semantic Segmentationcited by 9
  29. 2024Registration by Regression (RbR): a framework for interpretable and flexible atlas registrationcited by 8
  30. 2024Hierarchical Uncertainty Estimation for Learning-based Registration in Neuroimagingcited by 4
  31. 2024Learn2Synth: Learning Optimal Data Synthesis Using Hypergradients for Brain Image Segmentationcited by 4

Built on by

Not cited yet.