CVPR Findings2026
TP-Seg: Task-Prototype Framework for Unified Medical Lesion Segmentation
Jiawei Xu, Qiangqiang Zhou, Dandan Zhu, Yong Chen, Yugen Yi, Xiaoqi Zhao
Abstract
Building a unified model with a single set of parameters to efficiently handle diverse types of medical lesion segmentation has become a crucial objective for AI-assisted diagnosis. Existing unified segmentation approaches typically rely on shared encoders across heterogeneous tasks and modalities, which often leads to feature entanglement, gradient interference, and suboptimal lesion discrimination. In this work, we propose TP-Seg, a task-prototype framework for unified medical lesion segmentation. On one hand, the taskconditioned adapter effectively balances shared and taskspecific representations through a dual-path expert structure, enabling adaptive feature extraction across diverse medical imaging modalities and lesion types. On the other hand, the prototype-guided task decoder introduces learnable task prototypes as semantic anchors and employs a cross-attention mechanism to achieve fine-grained modeling of task-specific foreground and background semantics. Without bells and whistles, TP-Seg consistently outperforms specialized, general and unified segmentation methods across 8 different medical lesion segmentation tasks covering multiple imaging modalities, demonstrating strong generalization, scalability and clinical applicability. Code and pretrained models will be available at the link
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
Builds on10
- LoRA: Low-Rank Adaptation of Large Language ModelsEdward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu et al.ICLR 2022 · 18,833 citations
- Segment AnythingAlexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao et al.ICCV 2023 · 13,211 citations
- Vision Transformer Adapter for Dense PredictionsZhe Chen, Yuchen Duan, Wenhai Wang, Junjun He et al.ICLR 2023 · 204 citations
- SegGPT: Towards Segmenting Everything In ContextXinlong Wang, Xiaosong Zhang, Yue Cao, Wen Wang et al.ICCV 2023 · 188 citations
- UniverSeg: Universal Medical Image SegmentationVictor Ion Butoi, Jose Javier Gonzalez Ortiz, Tianyu Ma, Mert R. Sabuncu et al.ICCV 2023 · 163 citations
Related papers
- Training Like a Medical Resident: Context-Prior Learning Toward Universal Medical Image SegmentationYunhe GaoCVPR 2024
- Bridging Radiology and Pathology Foundation Models via Concept-Based Multimodal Co-AdaptationYihang Chen, Yanyan Huang, Fuying Wang, Maximus Yeung et al.ICLR 2026
- MedM2G: Unifying Medical Multi-Modal Generation via Cross-Guided Diffusion with Visual InvariantChenlu Zhan, Yu Lin, Gaoang Wang, Hongwei Wang et al.CVPR 2024 · 20 citations
- Spider: A Unified Framework for Context-dependent Concept SegmentationXiaoqi Zhao, Youwei Pang, Wei Ji, Baicheng Sheng et al.ICML 2024 · 21 citations
- What Can Be Transferred: Unsupervised Domain Adaptation for Endoscopic Lesions SegmentationJiahua Dong, Yang Cong, Gan Sun, Bineng Zhong et al.CVPR 2020
