Main page for the DreamSim Project

This article explores how re-ranking methods enhance retrieval recall across anatomical structures in AI models. By applying re-ranking, all evaluated models—DreamSim, DINOv1, and SwinTransformer—show improved performance. While DreamSim consistently achieves the best results in region-based and localized retrieval, DINOv1 and SwinTransformer also excel in specific conditions. The findings highlight how re-ranking not only raises recall rates but also strengthens localization, proving its critical role in medical imaging and anatomical AI systems.

This article explores how different embedding approaches perform in medical image retrieval tasks. Self-supervised models slightly edge out supervised ones, though the performance gap across architectures is narrow. Surprisingly, pretraining on natural images (ImageNet) outperforms domain-specific sets (RadImageNet), while fractal-based embeddings achieve unexpectedly strong results given their synthetic origins. DreamSim, an ensemble of ViT embeddings fine-tuned with synthetic data, delivers the best recall overall, making it the current leader in embedding generation. Isolated anomalies—like poor recall for certain anatomies—remain unexplained, pointing to fertile ground for future research.