文献速递神经技术与安全CT

深度学习用于标准化头部CT重建:图像质量与操作者变异性的定量分析

Deep Learning for Standardized Head CT Reformatting: A Quantitative Analysis of Image Quality and Operator Variability.
PubMed 文献 · J Am Coll Radiol · 2025-09-23

研究用基础人工智能模型对1763例连续平扫头部CT进行自动重建,先在100例子集以专家手动标注验证模型准确性,再以验证后的模型为参考标准评估全数据集技术员原始重建的质量与速度。AI重建与专家标注高度一致,平均标志点定位误差0.6至0.9 mm;AI重建平均旋转误差0.7°、居中误差0.3%、缩放误差0.4%,而技术员重建分别为11.2°、6.4%、6.2%;手动重建质量在患者年龄、扫描仪位置、报告所见和个体技术员操作者等因素间存在显著变异。结论认为手动头部CT重建质量和速度变异大,单次深度学习基础模型可高精度、高一致性地生成重建,有望改善标准化、提升工作流效率并降低运营成本。

要点

  • AI自动重建与专家标注高度一致,平均标志点定位误差0.6至0.9 mm。
  • 技术员手动重建平均旋转误差11.2°、居中误差6.4%、缩放误差6.2%,远高于AI。
  • 手动重建质量在患者年龄、扫描仪位置、报告所见和个体技术员间存在显著变异。

核心结论

  • 手动头部CT重建在质量和速度上存在 substantial 变异,单次深度学习基础模型可高精度、高一致性地生成重建。
  • 实施此类自动化方法有望改善标准化、提升工作流效率并降低临床实践中的运营成本。

研究设计与样本

  • 研究使用基础人工智能模型为1763例连续平扫头部CT检查创建自动重建。
  • 模型准确性先在100例检查子集上验证,评估标志点检测以及旋转、居中和缩放误差,并与专家手动标注对照。
  • 验证后的模型随后作为参考标准,评估全数据集原始技术员生成重建的质量和速度。

诊断标准与关键数值

  • AI模型与专家标注高度一致,平均标志点定位误差0.6至0.9 mm。
  • 与专家定义平面相比,AI生成重建平均旋转误差0.7°、平均居中误差0.3%、平均缩放误差0.4%。
  • 技术员生成重建平均旋转误差11.2°、平均居中误差6.4%、平均缩放误差6.2%。

临床应用与阅片要点

  • 原文结论认为手动头部CT重建质量和速度变异大,单次深度学习基础模型可高精度、高一致性地生成重建。
  • 原文提出实施此类自动化方法有望改善标准化、提升工作流效率并降低运营成本。

局限性

  • 手动重建质量在患者年龄、扫描仪位置、报告所见和个体技术员操作者等因素间存在显著变异。

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 validate a deep learning foundation model for automated head CT reformatting and to quantify the quality, speed, and variability of conventional manual reformats in a real-world dataset.

METHODS: A foundation artificial intelligence (AI) model was used to create automated reformats for 1,763 consecutive noncontrast head CT examinations. Model accuracy was first validated on a 100-examination subset by assessing landmark detection as well as rotational, centering, and zoom error against expert manual annotations. The validated model was subsequently used as a reference standard to evaluate the quality and speed of the original technician-generated reformats from the full dataset.

RESULTS: The AI model demonstrated high concordance with expert annotations, with a mean landmark localization error of 0.6 to 0.9 mm. Compared with expert-defined planes, AI-generated reformats exhibited a mean rotational error of 0.7°, a mean centering error of 0.3%, and a mean zoom error of 0.4%. By contrast, technician-generated reformats demonstrated a mean rotational error of 11.2°, a mean centering error of 6.4%, and a mean zoom error of 6.2%. Significant variability in manual reformat quality was observed across different factors including patient age, scanner location, report findings, and individual technician operators.

CONCLUSION: Manual head CT reformatting is subject to substantial variability in both quality and speed. A single-shot deep learning foundation model can generate reformats with high accuracy and consistency. The implementation of such an automated method offers the potential to improve standardization, increase workflow efficiency, and reduce operational costs in clinical practice.

来源:J Am Coll Radiol,查看原文。版权归原作者所有。

引用

  1. Chang PD, Chu E, Floriolli D, et al. Deep Learning for Standardized Head CT Reformatting: A Quantitative Analysis of Image Quality and Operator Variability. J Am Coll Radiol. 2025. PMID: 40998241. doi:10.1016/j.jacr.2025.09.016

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。