用嵌入聚类优化医学图像分割冷启动，提升标注效率与精度。
From Cold Start to Active Learning: Embedding-Based Scan Selection for Medical Image Segmentation
- 基于基础模型嵌入与聚类自动选样本，构建多样初始训练集。
- 冷启动后结合不确定性与空间多样性，使Dice提升至0.939，Hausdorff距离降至19.16mm。
- 方法可解释且可视化，适合医疗数据少、标注成本高的场景。
准确的分割标注对疾病监测至关重要，但人工标注耗时且需专业知识。主动学习（AL）通过优先选择信息量大的样本减轻负担，通常分多样性驱动的冷启动阶段和不确定性驱动的选择阶段。本文提出一种新的冷启动采样策略，融合基础模型嵌入与聚类，自动确定聚类数量并按比例采样，构建多样且具代表性的初始训练集。随后采用结合空间多样性的不确定性主动学习框架进行样本选择。该方法直观可解释，支持候选样本特征空间分布的可视化。在涵盖X射线和MRI模态的三个数据集上评估：在CheXmask数据集上，冷启动使Dice从0.918升至0.929，Hausdorff距离从32.41降至27.66mm；AL阶段中，联合熵与多样性选择使Dice达0.939，Hausdorff距离降至19.16mm。Montgomery数据集上，冷启动使Dice从0.928增至0.950，距离从14.22降至9.38mm。SynthStrip上，冷启动小幅提升Dice，Hausdorff距离从9.43降至8.69mm；AL使Dice从0.816升至0.826，距离从7.76降至6.38mm。整体在低数据条件下显著优于基线方法，提升分割准确性。
原文摘要 · Abstract (English)
Accurate segmentation annotations are critical for disease monitoring, yet manual labeling remains a major bottleneck due to the time and expertise required. Active learning (AL) alleviates this burden by prioritizing informative samples for annotation, typically through a diversity-based cold-start phase followed by uncertainty-driven selection. We propose a novel cold-start sampling strategy that combines foundation-model embeddings with clustering, including automatic selection of the number of clusters and proportional sampling across clusters, to construct a diverse and representative initial training. This is followed by an uncertainty-based AL framework that integrates spatial diversity to guide sample selection. The proposed method is intuitive and interpretable, enabling visualization of the feature-space distribution of candidate samples. We evaluate our approach on three datasets spanning X-ray and MRI modalities. On the CheXmask dataset, the cold-start strategy outperforms random selection, improving Dice from 0.918 to 0.929 and reducing the Hausdorff distance from 32.41 to 27.66 mm. In the AL setting, combined entropy and diversity selection improves Dice from 0.919 to 0.939 and reduces the Hausdorff distance from 30.10 to 19.16 mm. On the Montgomery dataset, cold-start gains are substantial, with Dice improving from 0.928 to 0.950 and Hausdorff distance decreasing from 14.22 to 9.38 mm. On the SynthStrip dataset, cold-start selection slightly affects Dice but reduces the Hausdorff distance from 9.43 to 8.69 mm, while active learning improves Dice from 0.816 to 0.826 and reduces the Hausdorff distance from 7.76 to 6.38 mm. Overall, the proposed framework consistently outperforms baseline methods in low-data regimes, improving segmentation accuracy.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。