无需重训练，用流模型检测医学影像异常数据。
Safeguarding AI in Medical Imaging: Post-Hoc Out-of-Distribution Detection with Normalizing Flows
- 基于归一化流的后处理方法，不修改预训练模型权重。
- 在自建数据集上达84.61% AUROC，优于现有方法。
- 适合临床环境部署，保障AI诊断可靠性。
在人工智能驱动的医学影像中，无法检测分布外（OOD）数据会严重威胁临床可靠性，可能导致关键诊断错误。现有方法通常需要不切实际的重新训练或修改预训练模型，阻碍其在受监管临床环境中的应用。为此，我们提出一种无需改动预训练模型权重的后处理归一化流方法。我们在自建的、涵盖临床相关分布偏移的MedOOD数据集以及MedMNIST基准上进行了评估。该方法在MedOOD上达到84.61% AUROC，优于ViM（80.65%）和MDS（80.87%）；在MedMNIST上达到93.8% AUROC，超过ViM（88.08%）和ReAct（87.05%）。强性能与无需重训练的特性使其成为临床影像工作流中实用有效的安全防护方案。模型与构建数据集的代码已公开于https://github.com/dlotfi/MedOODFlow。
原文摘要 · Abstract (English)
In AI-driven medical imaging, the failure to detect out-of-distribution (OOD) data poses a severe risk to clinical reliability, potentially leading to critical diagnostic errors. Current OOD detection methods often demand impractical retraining or modifications to pre-trained models, hindering their adoption in regulated clinical environments. To address this challenge, we propose a post-hoc normalizing flow-based approach that seamlessly integrates with existing pre-trained models without altering their weights. We evaluate the approach on our in-house-curated MedOOD dataset, designed to capture clinically relevant distribution shifts, and on the MedMNIST benchmark. The proposed method achieves an AUROC of 84.61% on MedOOD, outperforming ViM (80.65%) and MDS (80.87%), and reaches 93.8% AUROC on MedMNIST, surpassing ViM (88.08%) and ReAct (87.05%). This combination of strong performance and post-hoc integration capability makes our approach a practical and effective safeguard for clinical imaging workflows. The model and code to build OOD datasets are publicly accessible at https://github.com/dlotfi/MedOODFlow.
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。