文献速递胸部CT

CT采集参数对基于RetinaNet的肺结节检测的影响:一项物理引导的验证研究

Effect of CT Acquisition Parameters on RetinaNet-Based Lung Nodule Detection: A Physics-Guided Validation Study Across Dose, Slice Thickness, and Reconstruction Kernel.
PubMed 文献 · Acad Radiol · 2026-07-23

该研究基于LIDC-IDRI数据集154例CT,使用MONAI RetinaNet模型(LUNA16预训练,未微调)评估六种采集条件对肺结节检测敏感度的影响。结论认为层厚和软核重建对AI检测性能的约束强于图像噪声。

要点

  • 软核重建和5mm层厚使AI肺结节检测敏感度下降超过10个百分点,需在QA中重点关注。
  • 25%剂量降低导致敏感度下降8.8个百分点,50%剂量降低仅下降4.8个百分点,呈不对称退化。
  • 3mm层厚对敏感度影响较小(下降2.4个百分点),但仍具统计学显著性。

核心结论

  • 在测试条件下,层厚和软核重建对AI检测性能的约束强于图像噪声。
  • 所提框架可重复、无需专有扫描仪数据,为部署后采集QA提供实用基础。

研究设计与样本

  • 回顾性验证研究,使用LIDC-IDRI数据集154例CT。
  • 采用MONAI RetinaNet模型,在LUNA16上预训练(fold 0),未微调。
  • 评估六种成像条件:基线、25%剂量降低、50%剂量降低、3mm层厚、5mm层厚、软核重建。
  • 剂量降低通过图像域高斯噪声模拟,层厚通过z轴移动平均模拟。
  • 检测敏感度在置信度阈值0.5、匹配标准15mm下计算。

诊断标准与关键数值

  • 基线敏感度为84.8%(95% CI: 80.2-89.8%)。
  • 软核重建敏感度74.3%(下降10.5个百分点,p < 0.05)。
  • 5mm层厚敏感度71.6%(下降13.2个百分点,p < 0.05)。
  • 25%剂量降低敏感度76.0%(下降8.8个百分点,p < 0.05),50%剂量降低敏感度80.0%(下降4.8个百分点,p < 0.05)。
  • 3mm层厚敏感度下降2.4个百分点(p = 0.009)。

临床应用与阅片要点

  • 日常阅片或AI辅助诊断报告中,应关注扫描层厚和重建核的选择对AI检出敏感度的影响。
  • 对于软核重建或5mm层厚图像,AI检测敏感度下降明显,需结合人工阅片复核。
  • 剂量降低引起的敏感度下降呈不对称性,25%剂量降低时下降更显著,需注意扫描协议优化。

局限性

  • 研究为回顾性设计,使用公开数据集LIDC-IDRI,可能限制泛化性。
  • 剂量降低和层厚变化通过模拟实现,非真实采集参数变化。
  • 模型未针对评估条件进行微调,可能影响性能表现。

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: Computed tomography acquisition parameters, including dose level, slice thickness, and reconstruction kernel, vary substantially across clinical sites, yet artificial intelligence (AI)-assisted lung nodule detection systems are rarely evaluated against this variability. We propose and demonstrate a physics-guided framework for evaluating detection model sensitivity to systematic acquisition parameter variation.

MATERIALS AND METHODS: 154 cases from the LIDC-IDRI dataset were evaluated using a Medical Open Network for AI (MONAI) RetinaNet model pretrained on LUNA16 (fold 0, no fine-tuning) across six imaging conditions: baseline, 25% dose reduction, 50% dose reduction, 3 mm slice thickness, 5 mm slice thickness, and soft kernel reconstruction. Dose reduction was simulated via image-domain Gaussian noise; slice thickness via z-axis moving average. Detection sensitivity was computed at confidence threshold 0.5 with a 15 mm matching criterion.

RESULTS: Baseline sensitivity was 84.8% (95% CI: 80.2-89.8%). Soft kernel and 5 mm slice thickness produced the largest decreases: 74.3% (-10.5 pp, p < 0.05) and 71.6% (-13.2 pp, p < 0.05), respectively. Dose reduction caused moderate asymmetric degradation: 76.0% at 25% dose (-8.8 pp, p < 0.05) and 80.0% at 50% dose (-4.8 pp, p < 0.05). 3 mm slice thickness produced a small but significant reduction (-2.4 pp, p = 0.009). Reductions were most pronounced in the 3-6 mm range and consistent across confidence thresholds.

CONCLUSION: Slice thickness and soft reconstruction kernel represent stronger constraints on AI detection performance than image noise under these conditions. The proposed framework is reproducible, requires no proprietary scanner data, and provides a practical basis for post-deployment acquisition QA.

来源:Acad Radiol,查看原文。版权归原作者所有。

引用

  1. Soliman D. Effect of CT Acquisition Parameters on RetinaNet-Based Lung Nodule Detection: A Physics-Guided Validation Study Across Dose, Slice Thickness, and Reconstruction Kernel. Acad Radiol. 2026. PMID: 42493351. doi:10.1016/j.acra.2026.06.056

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。