Pretraining a ResNet-18 with a Jaccard-weighted multi-label contrastive loss on modality and anatomy metadata improves fine-tuned AUC on two of three small medical imaging tasks.
Vis-mae: An efficient self-supervised learning approach on medical image segmentation and classification
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
eess.IV 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
ModAn-MulSupCon: Modality-and Anatomy-Aware Multi-Label Supervised Contrastive Pretraining for Medical Imaging
Pretraining a ResNet-18 with a Jaccard-weighted multi-label contrastive loss on modality and anatomy metadata improves fine-tuned AUC on two of three small medical imaging tasks.