Research map / Detection and segmentation / Open-set and few-shot detection
DeepIGeoS: A Deep Interactive Geodesic Framework for Medical Image Segmentation
Guotai Wang, María A. Zuluaga, Wenqi Li, Rosalind Pratt, Premal Amrishkumar Patel, Michaël Aertsen, Tom Doel, Anna Louise David, Jan A. Deprest, Sebastien Ourselin A, Tom K. Vercauteren
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 learned freeform pairwise CRF-Net(f) to P-Net | 2D P-Net · 2D placenta MRI (automatic) | Dice (%): 84.78 -> 85.86 | Improved |
| add Dense CRF post-processing to P-Net | 2D P-Net · 2D placenta MRI (automatic) | Dice (%): 84.78 -> 84.90 | No clear effect |
| use freeform instead of Gaussian pairwise potentials in CRF-Net (ablation) | 2D P-Net + CRF · 2D placenta MRI (automatic) | Dice (%): 85.44 -> 85.86 | Improved |
| use resolution-preserving P-Net instead of 5-block variant (ablation) | 2D P-Net · 2D placenta MRI (automatic) | Dice (%): 83.16 -> 84.78 | Improved |
| use geodesic instead of Euclidean distance in R-Net (ablation) | 2D R-Net · 2D placenta MRI (interactive refinement, first iteration) | Dice (%): 88.26 -> 88.76 | Improved |
| add CRF-Net(fu) with user hard constraints to R-Net (ablation) | 2D R-Net · 2D placenta MRI (interactive refinement, first iteration) | Dice (%): 88.76 -> 89.31 | Improved |
| use geodesic instead of Euclidean distance when CRF-Net(fu) is used (ablation) | 2D R-Net + CRF-Net(fu) · 2D placenta MRI (interactive refinement, first iteration) | Dice (%): 88.71 -> 89.31 | Improved |
| use R-Net instead of min-cut user-editing for refinement | 2D refinement · 2D placenta MRI (interactive refinement, first iteration) | Dice (%): 87.04 -> 88.76 | Improved |
| use resolution-preserving P-Net instead of 5-block variant (ablation) | 3D P-Net · BraTS 2015 FLAIR whole tumor (automatic) | Dice (%): 85.36 -> 86.68 | Improved |
| add Dense CRF to 3D P-Net | 3D P-Net · BraTS 2015 FLAIR whole tumor (automatic) | Dice (%): 86.68 -> 87.06 | No clear effect |
| add CRF-Net(f) to 3D P-Net | 3D P-Net · BraTS 2015 FLAIR whole tumor (automatic) | Dice (%): 86.68 -> 87.55 | Improved |
| use geodesic instead of Euclidean distance in 3D R-Net (ablation) | 3D R-Net · BraTS 2015 FLAIR whole tumor (interactive refinement, first iteration) | Dice (%): 88.82 -> 89.30 | Improved |
Closest papers by meaning
- Deep Interactive Segmentation of Medical Images: A Systematic Review and Taxonomy (TPAMI 2024 · similarity 0.89)
- Learning Topological Interactions for Multi-Class Medical Image Segmentation (ECCV 2022 · similarity 0.86)
- Interactive Medical Image Segmentation: A Benchmark Dataset and Baseline (CVPR 2025 · similarity 0.86)
- Generative Medical Segmentation (AAAI 2025 · similarity 0.86)
- UniPro: Unified Multi-Mode Medical Image Segmentation from 2D Images to 3D Volumes via Propagation (NeurIPS 2026 · similarity 0.85)
- Combinatorial CNN-Transformer Learning with Manifold Constraints for Semi-supervised Medical Image Segmentation (AAAI 2024 · 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.
