Research map / Label-efficient and robust learning / Self-supervised and contrastive
3D Self-Supervised Methods for Medical Imaging
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Pretrain with a 3D self-supervised task rather than train from scratch | 3D segmentation models · BraTS 2018 and pancreas tumor segmentation | Segmentation performance | Helped |
| Use full 3D context rather than pretraining on 2D slices | Self-supervised pretraining for 3D segmentation · BraTS 2018 and pancreas tumor segmentation | Segmentation performance | Helped |
| Pretrain with 3D-RPL | 3D U-Net · BraTS 2018 validation | Dice (ET and TC) | Helped |
| Pretrain on a large unlabeled UK Biobank corpus, then fine-tune on the downstream dataset | 2D self-supervised methods · Diabetic Retinopathy 2019 | Quadratic Weighted Kappa | Helped |
| Use self-supervised pretraining instead of training from scratch | 3D segmentation models · Pancreas tumor segmentation | Speed of convergence | Helped |
| Use self-supervised pretraining instead of training from scratch | 2D classification models · Diabetic Retinopathy 2019 | Speed of convergence | Helped |
| Freeze the pretrained encoder for initial warm-up epochs, then fine-tune it (ablation) | Pretrained 3D segmentation models · 3D segmentation tasks | Downstream performance | Helped |
| Fine-tune the encoder immediately with a randomly initialized decoder (ablation) | Pretrained 3D segmentation models · 3D segmentation tasks | Downstream performance | Hurt |
| Keep the encoder frozen throughout training (ablation) | Pretrained models · Downstream tasks | Downstream performance | Hurt |
| Include each of the listed augmentations when forming Exemplar positives (ablation) | 3D-Exemplar · Downstream tasks | Downstream performance | Helped |
Closest papers by meaning
- Revisiting MAE Pre-training for 3D Medical Image Segmentation (CVPR 2025 · similarity 0.91)
- Autoregressive Sequence Modeling for 3D Medical Image Representation (AAAI 2025 · similarity 0.88)
- Masked Scene Modeling: Narrowing the Gap Between Supervised and Self-Supervised Learning in 3D Scene Understanding (CVPR 2025 · similarity 0.86)
- Common3D: Self-Supervised Learning of 3D Morphable Models for Common Objects in Neural Feature Space (CVPR 2025 · similarity 0.86)
- How Well Do Supervised 3D Models Transfer to Medical Imaging Tasks? (ICLR 2024 · similarity 0.86)
- Integrating Deep Metric Learning with Coreset for Active Learning in 3D Segmentation (NeurIPS 2024 · 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.