文献速递技术与安全胸部腹部CT

深度学习三维摄像头对CT检查工作流程和患者定位的影响

Impact of the use of deep-learning 3D camera on workflow and patient positioning in CT examinations.
PubMed 文献 · Eur J Radiol · 2025-07-17

前瞻性研究分析596例接受非增强胸腹盆CT检查的患者,比较手动定位与深度学习三维摄像头自动定位。摄像头组CT检查室时间中位数短于手动组(223秒 vs 255秒,P<0.001),但两组偏心距离、CTDIvol、DLP及背景噪声无显著差异。结论认为摄像头定位改善了工作流程,但对定位精度和辐射剂量的影响极小。

要点

  • 深度学习三维摄像头自动定位可缩短CT检查室时间(中位数223秒 vs 255秒,P<0.001)。
  • 摄像头组与手动组偏心距离无显著差异(中位数14 mm vs 13 mm,P=0.30)。
  • 两组CTDIvol、DLP及背景噪声无显著差异(P值分别为0.58、0.64、0.19)。

核心结论

  • 深度学习三维摄像头定位改善了CT检查工作流程,但对偏心距离、CTDIvol、DLP和背景噪声的影响极小。

研究设计与样本

  • 前瞻性研究,纳入596例受试者(中位年龄71岁,364名男性),于2023年10月至2024年1月接受非增强胸腹盆CT检查,分为手动定位组和深度学习三维摄像头自动定位组,两组扫描参数完全匹配。

诊断标准与关键数值

  • CT检查室时间(从进入至离开CT扫描室)中位数:摄像头组223秒 vs 手动组255秒,P<0.001。
  • 偏心距离(患者与扫描器等中心之差)中位数:摄像头组14 mm vs 手动组13 mm,P=0.30。
  • CTDIvol中位数:摄像头组5 mGy vs 手动组6 mGy,P=0.58。
  • DLP中位数:摄像头组410 mGy*cm vs 手动组416 mGy*cm,P=0.64。
  • 背景噪声(扫描范围中点腹主动脉CT衰减的标准差)中位数:摄像头组9 HU vs 手动组9 HU,P=0.19。

临床应用与阅片要点

  • 深度学习三维摄像头自动定位可用于缩短CT检查室时间,但原文未据此给出具体操作推荐。

局限性

  • 原文未报告局限性。

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: This study aims to assess the impact of the use of deep-learning three-dimensional (3D) camera on workflow and patient positioning in CT examinations in comparison with manual positioning by radiographers.

METHODS: This prospective study analyzed 596 participants (median age, 71 years; 364 men) who underwent unenhanced chest-abdomen-pelvis CT between October 2023 and January 2024 by either manual positioning (manual group) or automatic positioning using a deep-learning 3D camera (camera group). The scan parameters used in the two groups were completely matched. The CT room time, from entering to leaving the CT scan room, was compared between the two groups. Moreover, the off-center distance, which is the difference between the patient and scanner's isocenters, CT dose-index volume (CTDIvol), dose-length product (DLP), and standard deviation (SD) of the CT attenuation of the abdominal aorta at the midpoint of the scan range as background noise were compared between the two groups.

RESULTS: The median CT room time was shorter in the camera group than in the manual group (223 s vs. 255 s; P < 0.001). No difference in the median off-center distance (14 mm vs. 13 mm; P = 0.30), CTDIvol (5 mGy vs. 6 mGy; P = 0.58), DLP (410 mGy*cm vs. 416 mGy*cm; P = 0.64), and background noise (9 HU vs. 9 HU; P = 0.19) was observed between the manual and camera groups.

CONCLUSION: Patient positioning with a deep-learning 3D camera improved the workflow of CT examinations compared with manual positioning by radiographers. However, its impacts on off-center distance, CTDIvol, DLP, and background noise were minimal.

来源:Eur J Radiol,查看原文。版权归原作者所有。

引用

  1. Ito A, Noda Y, Kawai N, et al. Impact of the use of deep-learning 3D camera on workflow and patient positioning in CT examinations. Eur J Radiol. 2025. PMID: 40729987. doi:10.1016/j.ejrad.2025.112324

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。