Research map / Label-efficient and robust learning / Self-supervised and contrastive
Modeling the Density of Pixel-level Self-supervised Embeddings for Unsupervised Pathology Segmentation in Medical CT
Mikhail Goncharov, Eugenia Soboleva, Daniil Ignatyev, Mariia Donskova, Mikhail Belyaev, Ivan Oseledets, Marina Munkhoeva, Maxim Panov
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 masking-invariant condition model (vs no condition) to Gaussian density model (ablation) | Screener Gaussian density · KiTS | voxel-level AUROC: 0.61 -> 0.87 | Improved |
| add masking-invariant condition model (vs no condition) to Gaussian density model (ablation) | Screener Gaussian density · LIDC | voxel-level AUROC: 0.81 -> 0.96 | Improved |
| replace sin-cos positional encoding with masking-invariant condition (Gaussian) (ablation) | Screener Gaussian density · LiTS | voxel-level AUROC: 0.77 -> 0.90 | Improved |
| replace APE with masking-invariant condition (Gaussian) (ablation) | Screener Gaussian density · LiTS | voxel-level AUROC: 0.86 -> 0.90 | Improved |
| add conditioning to normalizing flow density model (ablation) | Screener flow density · LIDC | voxel-level AUROC: 0.96 -> 0.96 | No clear effect |
| replace ImageNet-pretrained descriptor (MSFlow) with dense SSL descriptor (ablation) | density-based UVAS · MIDRC | voxel-level AUROC: 0.66 -> 0.89 | Improved |
| replace supervised STU-Net descriptor with DenseVICReg descriptor (ablation) | normalizing flow UVAS · MIDRC | voxel-level AUROC: 0.44 -> 0.89 | Improved |
| use DenseVICReg instead of DenseInfoNCE descriptor (ablation) | normalizing flow UVAS · LiTS | voxel-level AUROC: 0.91 -> 0.93 | Improved |
| increase descriptor dimension from 32 to 128 (ablation) | DenseVICReg descriptor · KiTS | voxel-level AUROC: 0.88 -> 0.87 | No clear effect |
| add Screener pretraining before fine-tuning (25 labeled cases) | nnUNet · LIDC | Dice: 0.21 -> 0.31 | Improved |
| add Screener pretraining before fine-tuning (25 labeled cases) | nnUNet · KiTS | Dice: 0.41 -> 0.43 | No clear effect |
| add Screener pretraining before fine-tuning on full data | nnUNet · KiTS | Dice: 0.78 -> 0.76 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- VoCo — VoCo: A Simple-yet-Effective Volume Contrastive Learning Framework for 3D Medical Image Analysis (CVPR 2024) · better on 2 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- PathVQ: Reforming Computational Pathology Foundation Model for Whole Slide Image Analysis via Vector Quantization (NeurIPS 2025 · similarity 0.85)
- Unraveling Normal Anatomy via Fluid-Driven Anomaly Randomization (CVPR 2025 · similarity 0.85)
- Advancing Generalizable Tumor Segmentation with Anomaly-Aware Open-Vocabulary Attention Maps and Frozen Foundation Diffusion Models (CVPR 2025 · similarity 0.85)
- Beyond ViT Tokens: Masked-Diffusion Pretrained Convolutional Pathology Foundation Model for Cell-Level Dense Prediction (NeurIPS 2026 · similarity 0.85)
- Prototype-Based Image Prompting for Weakly Supervised Histopathological Image Segmentation (CVPR 2025 · similarity 0.85)
- Neighbor Does Matter: Density-Aware Contrastive Learning for Medical Semi-supervised Segmentation (AAAI 2025 · similarity 0.85)
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.
