Research map / Detection and segmentation / Open-set and few-shot detection
Learning Topological Interactions for Multi-Class Medical Image Segmentation
Saumya Gupta, Xiaoling Hu, James Kaan, Michael Jin, Mutshipay Mpoy, Katherine Chung, Gagandeep Singh, Mary Saltz, Tahsin Kurc, Joel Saltz, Apostolos Tassiopoulos, Prateek Prasanna and 1 more
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add topological interaction loss L_ti (ablation) | nnUNet · IVUS lumen | Dice: 0.893 -> 0.949 | Improved |
| add topological interaction loss L_ti (ablation) | nnUNet · IVUS lumen | % violations: 2.708 -> 0.157 | Improved |
| add topological interaction loss L_ti (ablation) | nnUNet · IVUS media | Dice: 0.856 -> 0.910 | Improved |
| add topological interaction loss L_ti | nnUNet · Aorta lumen | % violations: 5.424 -> 3.526 | Improved |
| add topological interaction loss L_ti | nnUNet · Aorta wall | Dice: 0.741 -> 0.758 | Improved |
| add topological interaction loss L_ti | nnUNet · Multi-Atlas kidney | Dice: 0.931 -> 0.936 | No clear effect |
| add topological interaction loss L_ti | nnUNet · Multi-Atlas spleen | Dice: 0.950 -> 0.960 | Improved |
| add topological interaction loss L_ti | nnUNet · SegTHOR trachea | Dice: 0.910 -> 0.929 | Improved |
| use MSE instead of CE as pixel loss in L_ti (ablation) | Ours · IVUS lumen | Dice: 0.949 -> 0.915 | Hurt |
| use Dice loss instead of CE as pixel loss in L_ti (ablation) | Ours · IVUS lumen | Dice: 0.949 -> 0.937 | Hurt |
| lower topological loss weight to 5.0e-5 (ablation) | Ours · IVUS lumen | Dice: 0.949 -> 0.913 | Hurt |
| raise topological loss weight to 2.0e-4 (ablation) | Ours · IVUS lumen | Dice: 0.949 -> 0.938 | Hurt |
Closest papers by meaning
- TopoCL: Topological Contrastive Learning for Medical Imaging (CVPR 2026 · similarity 0.88)
- A Topological Loss Function for Deep-Learning Based Image Segmentation Using Persistent Homology (TPAMI 2020 · similarity 0.87)
- Conformable Convolution for Topologically Constrained Learning of Complex Anatomical Structures (AAAI 2026 · similarity 0.87)
- Deep Interactive Segmentation of Medical Images: A Systematic Review and Taxonomy (TPAMI 2024 · similarity 0.87)
- Combinatorial CNN-Transformer Learning with Manifold Constraints for Semi-supervised Medical Image Segmentation (AAAI 2024 · similarity 0.86)
- Topograph: An Efficient Graph-Based Framework for Strictly Topology Preserving Image Segmentation (ICLR 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
