… level representations particularly useful for segmentation tasks by … local contrastive loss
to learn useful local features. We demonstrate on three medical datasets for the segmentation …

… Unlike natural image datasets, for the evaluated medical imaging datasets we observe that
we may not require large batch sizes in the pre-training stage to obtain high performance. …

… for volumetric medical image segmentation with limited annotations. The first stage is feature
… To achieve this, for one local feature q , in addition to its local positives P ( q ) , we define …

… We propose a supervised local contrastive loss that leverages limited pixel-wise annotation
… -supervised contrastive learning scheme is used to learn both global and local features from …

… In this paper, we propose a local contrastive loss to learn good pixel level features useful for
… with limited labels. Unsupervised learning of good local features can be as crucial as global …

… performance, thereby improving segmentation performance with limited annotations. Exten…
medical image segmentation. We encourage more useful global and local features via multi-…

… contrastive learning to learn a prior extractor from the unlabeled query images to benefit the
few shot segmentation … for low-contrast medical image segmentation,” IEEE Transactions on …

… medical image segmentation with limited annotations. More specifically, we exchange the
features … To achieve this, for one local feature q, in addition to its local positives P(q), we define …

… tasks with limited labels. A critical step in … positional contrastive learning (PCL) framework
to generate contrastive data pairs by leveraging the position information in volumetric medical …

… Konukoglu, “Contrastive learning of global and local features for medical image
segmentation with limited annotations,” arXiv preprint arXiv:2006.10511, 2020. [37] X. Zhao, R. …