Research map / Detection and segmentation / Open-set and few-shot detection
FlowSDF: Flow Matching for Medical Image Segmentation Using Distance Transforms
Lea Bogensperger, Dominik Narnhofer, Alexander Falk, Konrad Schindler B, Thomas Pock
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 |
|---|---|---|---|
| represent masks as SDF instead of binary masks (ablation) | FlowSDF (K=1) · MoNuSeg | F1: 79.78 -> 80.67 | Improved |
| represent masks as SDF instead of binary masks (ablation) | FlowSDF (K=1) · MoNuSeg | mIoU: 66.45 -> 67.70 | Improved |
| represent masks as SDF instead of binary masks (ablation) | FlowSDF (K=1) · GlaS | F1: 88.20 -> 88.64 | Improved |
| represent masks as SDF instead of binary masks (ablation) | FlowSDF (K=1) · GlaS | mIoU: 79.91 -> 80.63 | Improved |
| add patch-averaging over overlapping patches (ablation) | FlowSDF · MoNuSeg | F1: 80.45 -> 80.67 | Improved |
| add patch-averaging over overlapping patches (ablation) | FlowSDF · MoNuSeg | mIoU: 67.38 -> 67.70 | Improved |
| average K=25 runs for MMSE (ablation) | FlowSDF · MoNuSeg | F1: 80.67 -> 81.02 | Improved |
| average K=25 runs for MMSE (ablation) | FlowSDF · GlaS | mIoU: 80.63 -> 81.05 | Improved |
| add T=100 noise injection steps (ablation) | FlowSDF MMSE (K=25) · GlaS | mIoU: 81.05 -> 81.71 | Improved |
| add T=100 noise injection steps (ablation) | FlowSDF MMSE (K=25) · MoNuSeg | F1: 81.02 -> 81.22 | Improved |
| increase K from 1 to 25 | SegDiff · MoNuSeg | F1: 76.88 -> 81.19 | Improved |
| use flow matching ODE with SDF instead of VE SDE with SDF | FlowSDF (K=1) · GlaS | mIoU: 71.36 -> 80.63 | Improved |
Closest papers by meaning
- MedFlowSeg: Flow Matching for Medical Image Segmentation with Frequency-Aware Attention (NeurIPS 2026 · similarity 0.89)
- STREAM: Stochastic Riemannian Flow Matching with Anisotropic Decoder for Digital Histopathology Image Generation (NeurIPS 2026 · similarity 0.84)
- ConDSeg: A General Medical Image Segmentation Framework via Contrast-Driven Feature Enhancement (AAAI 2025 · similarity 0.84)
- Rethinking Vector Field Learning for Generative Segmentation (NeurIPS 2026 · similarity 0.84)
- Improving Flow Matching by Aligning Flow Divergence (ICML 2025 · similarity 0.84)
- MedSegDiff-V2: Diffusion-Based Medical Image Segmentation with Transformer (AAAI 2024 · similarity 0.83)
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.
