用大规模无标注MRI数据预训练，提升脑转移瘤复发与放疗坏死的区分准确率。
Large-Scale Pre-training Enables Multimodal AI Differentiation of Radiation Necrosis from Brain Metastasis Progression on Routine MRI
- 先在1万+无标注MRI上自监督预训练视觉变换器，再用少量有标签数据微调。
- 在同中心和跨中心验证中，模型AUC达0.947和0.821，优于传统方法。
- 结合影像与分割图的多模态输入，结果更优且注意力可视化可解释。
背景：立体定向放射外科（SRS）后区分放疗坏死（RN）与肿瘤进展仍是脑转移瘤治疗中的关键挑战。虽组织病理学为金标准，但其侵入性限制了可行性。传统监督学习受限于稀缺的活检确诊数据。自监督学习（SSL）可通过利用日益丰富的大规模无标注脑转移瘤影像数据集克服此局限。方法：采用两阶段深度学习策略，受基础模型范式启发，使用10,167个来自多源的未标注T1CE MRI子体积对视觉变换器（ViT）进行自监督预训练。随后，在公开的MOLAB数据集（n=109）上，以双通道输入（T1CE MRI与分割掩码）对预训练ViT进行RN分类微调，其中20%数据作为同中心保留测试集。外部验证在另一中心测试队列（n=28）上进行。结果：自监督模型在同中心测试集中达到AUC 0.916，跨中心测试中为0.764，显著优于全监督ViT（AUC 0.624/0.496；p=0.001/0.008）和放射组学（AUC 0.807/0.691；p=0.005/0.014）。多模态融合进一步提升性能（AUC 0.947/0.821；p=0.073/0.001）。注意力图可视化显示模型聚焦于临床相关病灶区域。结论：在日益可用的无标注脑转移瘤数据集上进行大规模预训练，可显著提升AI模型性能。该两阶段多模态深度学习策略仅基于常规T1CE MRI与标准临床数据，即实现了高精度区分放疗坏死与肿瘤进展，提供了一种可解释、临床可及的解决方案，值得进一步验证。
原文摘要 · Abstract (English)
Background: Differentiating radiation necrosis (RN) from tumor progression after stereotactic radiosurgery (SRS) remains a critical challenge in brain metastases. While histopathology represents the gold standard, its invasiveness limits feasibility. Conventional supervised deep learning approaches are constrained by scarce biopsy-confirmed training data. Self-supervised learning (SSL) overcomes this by leveraging the growing availability of large-scale unlabeled brain metastases imaging datasets. Methods: In a two-phase deep learning strategy inspired by the foundation model paradigm, a Vision Transformer (ViT) was pre-trained via SSL on 10,167 unlabeled multi-source T1CE MRI sub-volumes. The pre-trained ViT was then fine-tuned for RN classification using a two-channel input (T1CE MRI and segmentation masks) on the public MOLAB dataset (n=109) using 20% of datasets as same-center held-out test set. External validation was performed on a second-center test cohort (n=28). Results: The self-supervised model achieved an AUC of 0.916 on the same-center test set and 0.764 on the second center test set, surpassing the fully supervised ViT (AUC 0.624/0.496; p=0.001/0.008) and radiomics (AUC 0.807/0.691; p=0.005/0.014). Multimodal integration further improved performance (AUC 0.947/0.821; p=0.073/0.001). Attention map visualizations enabled interpretability showing the model focused on clinically relevant lesion subregions. Conclusion: Large-scale pre-training on increasingly available unlabeled brain metastases datasets substantially improves AI model performance. A two-phase multimodal deep learning strategy achieved high accuracy in differentiating radiation necrosis from tumor progression using only routine T1CE MRI and standard clinical data, providing an interpretable, clinically accessible solution that warrants further validation.
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。