Self-Paced Contrastive Learning for Semi-supervised Medical Image Segmentation with Meta-labels
arXiv:2107.13741
Abstract
Pre-training a recognition model with contrastive learning on a large dataset of unlabeled data has shown great potential to boost the performance of a downstream task, e.g., image classification. However, in domains such as medical imaging, collecting unlabeled data can be challenging and expensive. In this work, we propose to adapt contrastive learning to work with meta-label annotations, for improving the model's performance in medical image segmentation even when no additional unlabeled data is available. Meta-labels such as the location of a 2D slice in a 3D MRI scan or the type of device used, often come for free during the acquisition process. We use the meta-labels for pre-training the image encoder as well as to regularize a semi-supervised training, in which a reduced set of annotated data is used for training. Finally, to fully exploit the weak annotations, a self-paced learning approach is used to help the learning and discriminate useful labels from noise. Results on three different medical image segmentation datasets show that our approach: i) highly boosts the performance of a model trained on a few scans, ii) outperforms previous contrastive and semi-supervised approaches, and iii) reaches close to the performance of a model trained on the full data.
References in corpus (19)
- mixup: Beyond Empirical Risk Minimization
- Improved Baselines with Momentum Contrastive Learning
- ADVENT: Adversarial Entropy Minimization for Domain Adaptation in Semantic Segmentation
- On the Variance of the Adaptive Learning Rate and Beyond
- Contrastive Multiview Coding
- A scoping review of transfer learning research on medical image analysis using ImageNet
- Big Self-Supervised Models are Strong Semi-Supervised Learners
- What makes ImageNet good for transfer learning?
- Semi-Supervised Medical Image Segmentation via Learning Consistency under Transformations
- The loss surface of deep and wide neural networks
- No bad local minima: Data independent training error guarantees for multilayer neural networks
- Supervised Contrastive Learning
- Adversarial Self-Supervised Contrastive Learning
- Bridging Saliency Detection to Weakly Supervised Object Detection Based on Self-paced Curriculum Learning
- Margin Preserving Self-paced Contrastive Learning Towards Domain Adaptation for Medical Image Segmentation
- Competence-based Curriculum Learning for Neural Machine Translation
- Deep Co-Training for Semi-Supervised Image Segmentation
- A Two-Stream Mutual Attention Network for Semi-supervised Biomedical Segmentation with Noisy Labels
- Curriculum Learning Strategies for IR: An Empirical Study on Conversation Response Ranking