文献速递妇产影像MR

基础模型赋能多模态预测高级别浆液性卵巢癌铂耐药

Foundation Model-empowered Multimodal Prediction of Platinum Resistance in High-grade Serous Ovarian Cancer: A Multicenter Retrospective Study.
PubMed 文献 · Acad Radiol · 2026-07-17

本研究为多中心回顾性队列研究,纳入744例接受初次减瘤手术的高级别浆液性卵巢癌患者,收集临床变量、术前盆腔MRI和术后H&E全切片图像。采用预训练放射学与病理学基础模型作为特征提取器,结合跨模态融合模块构建预测铂耐药模型。三模态模型在各队列AUC最高,范围为0.779至0.790,优于晚期融合基线模型。

要点

  • 三模态(临床+MRI+病理)模型预测铂耐药AUC最高,优于单模态及双模态。
  • 该框架支持缺失模态推断,对日常资料不全病例有潜在应用价值。
  • 模型性能在多中心外部验证中保持稳定,提示一定泛化能力。

核心结论

  • 基础模型赋能的多模态模型整合临床、MRI与病理数据可预测高级别浆液性卵巢癌铂耐药,跨中心性能稳定且优于晚期融合基线模型。

研究设计与样本

  • 多中心回顾性队列研究,纳入744例接受初次减瘤手术的高级别浆液性卵巢癌患者,来自四个中心。
  • 收集围手术期多模态数据,包括临床变量、术前盆腔MRI和术后H&E全切片图像。
  • 使用预训练放射学与病理学基础模型作为冻结特征提取器,采用基于注意力的聚合处理可变长度MRI切片和WSI图块。
  • 跨模态融合模块整合各模态并支持缺失模态推断,以受试者工作特征曲线下面积(AUC)评估性能。

诊断标准与关键数值

  • 内部验证队列中,单模态及多模态基础模型变体AUC范围为0.708至0.780。
  • 外部测试队列A、B、C中,AUC范围为0.653至0.790。
  • 三模态基础模型(临床+MRI+WSI)在各队列AUC最高:内部验证0.780,外部队列A、B、C分别为0.779、0.781、0.790。
  • 三模态晚期融合模型AUC:内部验证0.753,外部队列A、B、C分别为0.729、0.715、0.730。

临床应用与阅片要点

  • 临床、MRI与病理三模态信息联合可提升高级别浆液性卵巢癌铂耐药预测效能。
  • 框架支持缺失模态推断,可用于部分影像或病理资料不全的临床场景。

局限性

  • 本研究为回顾性设计,存在选择偏倚与信息偏倚可能。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

英文原文(摘要)· 点击展开

RATIONALE AND OBJECTIVES: To develop and validate a foundation model (FM)-empowered multimodal framework to predict platinum resistance in patients with high-grade serous ovarian cancer (HGSOC).

MATERIALS AND METHODS: This multicenter retrospective cohort study included 744 HGSOC patients who underwent primary debulking surgery across four centers. Perioperative multimodal data were collected, including clinical variables, preoperative pelvic magnetic resonance imaging (MRI), and postoperative hematoxylin and eosin (H&E)-stained whole slide images (WSIs). Pretrained radiology and pathology FMs were used as frozen feature extractors, and attention-based aggregation was applied to summarize variable-length MRI slices and WSI tiles. A cross-modal fusion module integrated modalities and supported missing-modality inference. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC) and compared with late-fusion (LF) baselines.

RESULTS: In the internal validation cohort, unimodal and multimodal FM variants achieved AUCs ranging from 0.708 to 0.780, and in external test cohorts A, B, and C achieved AUCs ranging from 0.653 to 0.790. The trimodal FM (clinical+MRI+WSI) achieved the highest AUC in all cohorts (internal validation: 0.780; external cohorts A, B, and C: 0.779, 0.781, and 0.790, respectively), outperforming the trimodal LF model (internal validation: 0.753; external cohorts A, B, and C: 0.729, 0.715, and 0.730, respectively).

CONCLUSION: A FM-empowered multimodal model integrating clinical, MRI, and pathological data was constructed to predict platinum resistance in HGSOC, demonstrating superior and stable performance across centers.

来源:Acad Radiol,查看原文。版权归原作者所有。

引用

  1. Bi, Qu, Liu, et al. Foundation Model-empowered Multimodal Prediction of Platinum Resistance in High-grade Serous Ovarian Cancer: A Multicenter Retrospective Study. Acad Radiol. 2026. PMID: 42469040. doi:10.1016/j.acra.2026.06.024

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。