M3Ret: Unleashing Zero-shot Multimodal Medical Image Retrieval via Self-Supervision
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866908512348536832 |
|---|---|
| author | Liu, Che Jiang, Zheng Fang, Chengyu Guo, Heng Zhou, Yan-Jie Qu, Jiaqi Lu, Le Xu, Minfeng |
| author_facet | Liu, Che Jiang, Zheng Fang, Chengyu Guo, Heng Zhou, Yan-Jie Qu, Jiaqi Lu, Le Xu, Minfeng |
| contents | Medical image retrieval is essential for clinical decision-making and translational research, relying on discriminative visual representations. Yet, current methods remain fragmented, relying on separate architectures and training strategies for 2D, 3D, and video-based medical data. This modality-specific design hampers scalability and inhibits the development of unified representations. To enable unified learning, we curate a large-scale hybrid-modality dataset comprising 867,653 medical imaging samples, including 2D X-rays and ultrasounds, RGB endoscopy videos, and 3D CT scans. Leveraging this dataset, we train M3Ret, a unified visual encoder without any modality-specific customization. It successfully learns transferable representations using both generative (MAE) and contrastive (SimDINO) self-supervised learning (SSL) paradigms. Our approach sets a new state-of-the-art in zero-shot image-to-image retrieval across all individual modalities, surpassing strong baselines such as DINOv3 and the text-supervised BMC-CLIP. More remarkably, strong cross-modal alignment emerges without paired data, and the model generalizes to unseen MRI tasks, despite never observing MRI during pretraining, demonstrating the generalizability of purely visual self-supervision to unseen modalities. Comprehensive analyses further validate the scalability of our framework across model and data sizes. These findings deliver a promising signal to the medical imaging community, positioning M3Ret as a step toward foundation models for visual SSL in multimodal medical image understanding. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_01360 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | M3Ret: Unleashing Zero-shot Multimodal Medical Image Retrieval via Self-Supervision Liu, Che Jiang, Zheng Fang, Chengyu Guo, Heng Zhou, Yan-Jie Qu, Jiaqi Lu, Le Xu, Minfeng Computer Vision and Pattern Recognition Machine Learning Medical image retrieval is essential for clinical decision-making and translational research, relying on discriminative visual representations. Yet, current methods remain fragmented, relying on separate architectures and training strategies for 2D, 3D, and video-based medical data. This modality-specific design hampers scalability and inhibits the development of unified representations. To enable unified learning, we curate a large-scale hybrid-modality dataset comprising 867,653 medical imaging samples, including 2D X-rays and ultrasounds, RGB endoscopy videos, and 3D CT scans. Leveraging this dataset, we train M3Ret, a unified visual encoder without any modality-specific customization. It successfully learns transferable representations using both generative (MAE) and contrastive (SimDINO) self-supervised learning (SSL) paradigms. Our approach sets a new state-of-the-art in zero-shot image-to-image retrieval across all individual modalities, surpassing strong baselines such as DINOv3 and the text-supervised BMC-CLIP. More remarkably, strong cross-modal alignment emerges without paired data, and the model generalizes to unseen MRI tasks, despite never observing MRI during pretraining, demonstrating the generalizability of purely visual self-supervision to unseen modalities. Comprehensive analyses further validate the scalability of our framework across model and data sizes. These findings deliver a promising signal to the medical imaging community, positioning M3Ret as a step toward foundation models for visual SSL in multimodal medical image understanding. |
| title | M3Ret: Unleashing Zero-shot Multimodal Medical Image Retrieval via Self-Supervision |
| topic | Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2509.01360 |