{"ID":6620738,"CreatedAt":"2026-07-15T01:01:48.440468303Z","UpdatedAt":"2026-07-15T03:28:55.185153975Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.12896","arxiv_id":"2607.12896","title":"UniMedSeg: Unified In-Context Learning for Multi-Paradigm 2D/3D Medical Image Segmentation","abstract":"Medical image segmentation foundation models are expected to generalize across diverse clinical scenarios, yet existing universal methods remain fragmented by prompt paradigms and spatial dimensions. Visual in-context learning, interactive segmentation, and language-guided segmentation are typically handled by paradigm-specific models, while 2D and 3D images are also modeled separately. Such isolation prevents heterogeneous annotations and data from being jointly absorbed by a single scalable model and limits cross-paradigm knowledge transfer. To address this bottleneck, we propose UniMedSeg, a Transformer-centric universal segmentation framework that maps visual examples, geometric interactions, language instructions, and 2D/3D images into a shared sequence space, enabling heterogeneous medical supervision to be jointly learned through a unified in-context interface without prompt- or dimension-specific branches. To overcome the long-sequence memory bottleneck caused by visual contexts, we introduce Decoupled Split Attention, which reduces attention complexity to linear while preserving hardware-friendly computation and focused context-target interaction. Extensively trained and evaluated on a large corpus curated from 27 public datasets, UniMedSeg achieves state-of-the-art performance across visual in-context, interactive, and language-guided segmentation without task-specific fine-tuning, demonstrating strong generalization on diverse held-out tasks. The code and model weights are publicly available at https://github.com/Lii1228/UniMedSeg","short_abstract":"Medical image segmentation foundation models are expected to generalize across diverse clinical scenarios, yet existing universal methods remain fragmented by prompt paradigms and spatial dimensions. Visual in-context learning, interactive segmentation, and language-guided segmentation are typically handled by paradigm...","url_abs":"https://arxiv.org/abs/2607.12896","url_pdf":"https://arxiv.org/pdf/2607.12896v1","authors":"[\"Yunzhou Li\",\"Jiesi Hu\",\"Yanwu Yang\",\"Hanyang Peng\",\"Chenfei Ye\",\"Jianfeng Cao\",\"Yixuan Yuan\",\"Ting Ma\"]","published":"2026-07-14T15:34:46Z","proceeding":"cs.CV","tasks":"[\"cs.CV\"]","methods":"[\"Transformer\"]","has_code":false,"code_links":[{"ID":614260,"CreatedAt":"2026-07-15T01:01:48.440468303Z","UpdatedAt":"2026-07-15T01:01:48.440468303Z","DeletedAt":null,"paper_id":6620738,"paper_url":"https://arxiv.org/abs/2607.12896","paper_title":"UniMedSeg: Unified In-Context Learning for Multi-Paradigm 2D/3D Medical Image Segmentation","repo_url":"https://github.com/Lii1228/UniMedSeg","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}