MRI深度学习自动分割头颈鳞癌颈部淋巴结的多中心验证
本研究基于两中心头颈MRI数据,对比两种淋巴结标注策略,构建并验证nnU-Net自动分割模型。CE-T1w结合边界贴合标注策略在内部验证队列获得最高Dice系数(0.795),外部测试队列证实了泛化能力。深度学习辅助显著缩短勾画时间33.14%(100.22秒 vs. 149.89秒,P<0.001),且不同标注策略对淋巴结体积测量产生显著差异(P<0.001)。
要点
- CE-T1w单序列模型即可实现颈部淋巴结自动3D分割,无需多序列输入
- 淋巴结勾画需明确标注策略,边界贴合与外缘缩进法体积结果差异显著
- 深度学习辅助勾画可减少约三分之一耗时,适合临床工作流整合
核心结论
- 基于CE-T1w参考预处理后单序列输入的nnU-Net模型能够准确高效地完成头颈鳞癌颈部淋巴结3D分割,并具备跨中心泛化能力。
- CE-T1w模型与手工分割具有高度体积一致性,边界贴合策略内外一致性相关系数为0.899,外缘缩进策略为0.939。
- 深度学习辅助勾画可显著缩短淋巴结标注时间,平均100.22秒对149.89秒,降幅33.14%。
研究设计与样本
- 回顾性多中心研究,收集两个中心头颈部MRI数据,内部验证队列99例,外部测试队列150例,另设851例用于评估勾画效率。
- 淋巴结采用两种手工标注策略:外缘缩进法(排除1-2mm淋巴结周围间隙)与边界贴合法(沿淋巴结轮廓勾画)。
- 使用nnU-Net框架分别训练CE-T1w、T1w、T2w三个单序列模型,以Dice相似系数和一致性相关系数评估分割与体积一致性。
关键影像表现/征象
- CE-T1w序列结合边界贴合标注策略训练模型在内部验证队列获得最高重叠精度,中位Dice相似系数为0.795。
- 两种标注策略下检出的淋巴结数量相近,但体积测量终点存在显著差异(P<0.001),提示标注标准对量化结果有实质影响。
诊断标准与关键数值
- 内部验证队列中CE-T1w结合边界贴合标注模型最高中位Dice相似系数为0.795。
- CE-T1w模型体积一致性方面,外缘缩进策略一致性相关系数为0.939,边界贴合策略为0.899。
- 深度学习辅助勾画平均耗时100.22秒,手工勾画平均耗时149.89秒,差异有统计学意义(P<0.001)。
临床应用与阅片要点
- 阅片报告涉及淋巴结体积随访时,应注明采用的勾画标准,避免因标注策略不同导致体积变化被误判。
- CE-T1w单序列即可支持自动分割流程,基层设备仅需常规增强T1序列即可部署,降低对多序列配准的依赖。
- 深度学习辅助分割结果建议人工复核,特别是淋巴结边缘与周围软组织分界不清的区域,以保障临床可用性。
局限性
- 研究为回顾性设计,难以完全排除选择偏倚对模型性能评估的影响。
- 外部验证仅覆盖两个中心的数据,对于设备厂商、磁场强度及扫描协议差异更大的多中心环境泛化性尚需进一步验证。
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
PURPOSE: To develop and externally validate an MRI-based deep learning framework for automated 3D segmentation of neck lymph nodes (LNs) in head and neck squamous cell carcinoma (HNSCC), and to assess segmentation accuracy, volumetric agreement, and time efficiency.
METHODS: Axial head and neck MRI data were retrospectively collected from two centers. Each LN was manually delineated following two strategies: margin-excluding approach (excluding a 1-2 mm perinodal margin) and boundary-adherent approach (following the LN contour). For each strategy, nnU-Net (a DL framework) was applied for training three single-sequence models using contrast-enhanced T1-weighted (CE-T1w), T1-weighted (T1w), and T2-weighted (T2w) imaging. The model performance was evaluated in the internal validation cohort (n = 99) and the external test cohort (n = 150). Efficiency was assessed by comparing manual vs. DL-assisted contouring times in a separate cohort (n = 851).
RESULTS: In the internal validation cohort, the model trained on CE-T1w images with boundary-adherent annotations achieved the highest overlap (median Dice similarity coefficient of 0.795). CE-T1w model showed high volumetric agreement with manual segmentation (concordance correlation coefficient of 0.939 for margin-excluding and 0.899 for boundary-adherent strategies), and its performance was confirmed in the external test cohort. DL assistance reduced contouring time by 33.14% (n = 851, mean 100.22 s vs. 149.89 s, p < 0.001). The two annotation policies yielded similar LN counts but significantly different volume endpoints (p < 0.001).
CONCLUSION: We developed and validated a deep learning framework capable of accurate and efficient 3D LN segmentation using single-sequence model inputs after CE-T1w-referenced preprocessing. The model shows strong generalization across multiple centers. This tool can standardize volumetric assessments and improve clinical workflow efficiency in HNSCC.
引用
- Sui, Qu, Xu, et al. Validation of MRI-based nnU-Net model for automated segmentation of neck lymph nodes in head and neck squamous cell carcinoma: a multicenter study. Neuroradiology. 2026. PMID: 42321574. doi:10.1007/s00234-026-04070-0
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。