Research map / Label-efficient and robust learning / Recognition and long-tail
Learning With Unsure Data for Medical Image Diagnosis
Botong Wu, Xinwei Sun, Lingjing Hu, Yizhou Wang
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 cost-sensitive parameters to UDM (ablation) | UDM · LIDC-IDRI | macro-F1: 69.34 -> 71.47 | Improved |
| add cost-sensitive parameters to UDM (ablation) | UDM · ADNI | macro-F1: 39.63 -> 40.91 | Improved |
| add conservative-aggressive strategy parameters to UDM+CS (ablation) | UDM+CS · LIDC-IDRI | F-beta macro (c-a): 71.47 -> 73.61 | Improved |
| add conservative-aggressive strategy parameters to UDM+CS (ablation) | UDM+CS · ADNI | F-beta macro (c-a): 38.38 -> 49.41 | Improved |
| add conservative-conservative strategy parameters to UDM+CS (ablation) | UDM+CS · LIDC-IDRI | F-beta macro (c-c): 67.87 -> 68.72 | Improved |
| add conservative-conservative strategy parameters to UDM+CS (ablation) | UDM+CS · ADNI | F-beta macro (c-c): 39.68 -> 42.03 | Improved |
| model unsure data as ordinal class (UDM) instead of cross-entropy | DenseNet-3D · LIDC-IDRI | macro-F1: 65.60 -> 69.34 | Improved |
| model unsure data as ordinal class (UDM) instead of cross-entropy | DenseNet-3D · ADNI | macro-F1: 32.62 -> 39.63 | Improved |
| train with unsure data instead of binary classifier without it | UDM+CS+CA · ADNI sure data | macro-F1: 79.13 -> 84.95 | Improved |
| train with unsure data instead of binary classifier without it | UDM+CS+CA · LIDC-IDRI sure data | macro-F1: 88.54 -> 88.92 | Improved |
| add cost-sensitive parameters to UDM (ablation) | UDM · LIDC-IDRI | number predicted as unsure: 467 -> 446 | Improved |
| add c-a strategy parameters (gamma) (ablation) | UDM+CS · LIDC-IDRI | number predicted positive: 153 -> 185 | Improved |
Closest papers by meaning
- Regretful Decisions under Label Noise (ICLR 2025 · similarity 0.84)
- Domain constraints improve risk prediction when outcome data is missing (ICLR 2024 · similarity 0.83)
- A Textbook Remedy for Domain Shifts: Knowledge Priors for Medical Image Analysis (NeurIPS 2024 · similarity 0.83)
- Novel Class Discovery in Chest X-rays via Paired Images and Text (AAAI 2024 · similarity 0.83)
- Medical thinking with multiple images (ICLR 2026 · similarity 0.83)
- SMART: Towards Pre-trained Missing-Aware Model for Patient Health Status Prediction (NeurIPS 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.
