Wu Lilan, Huang Shunfa, Xu Liling, Rao Shengxiang, Qian Zhen, Zhang Mengze, Yuan Ying, Zhou Jianjun
Department of Radiology, Zhongshan Hospital (Xiamen), Fudan University, Xiamen, China.
Xiamen Municipal Clinical Research Center for Medical Imaging, Xiamen, China.
Ther Adv Musculoskelet Dis. 2025 Sep 11;17:1759720X251374134. doi: 10.1177/1759720X251374134. eCollection 2025.
Since dual-energy x-ray absorptiometry (DXA) is currently the most commonly used reference standard, most previous studies using computed tomography (CT) attenuation values to predict osteoporosis have chosen abdominal CT images. A few studies have investigated whether the thoracic vertebrae can be independently used for the identification of osteoporosis compared to the lumbar vertebrae.
To investigate whether the attenuation values of thoracic vertebrae measured using artificial intelligence (AI) on chest CT would independently predict osteoporosis identification, considering central DXA as a reference standard.
This was a cross-sectional study.
A total of 553 participants (353 men and 200 women) who underwent chest CT and DXA within 1 day were included. The attenuation values (HU) of the T7-12 vertebrae and L1 vertebra were obtained by AI. The effects of the clinical baseline data and attenuation values among the normal, osteopenia, and osteoporosis groups were compared. The correlation between attenuation and bone mineral density (BMD) values was analyzed, and the diagnostic performance of thoracic and first lumbar vertebrae attenuation values for diagnosing osteopenia or osteoporosis was further explored.
The CT attenuation values of T7-12 and L1 vertebrae showed positive correlation with -score ( = 0.58-0.61, < 0.01). T12 attenuation >184.8 HU was 84.1% sensitive and 70.6% specific for distinguishing normal BMD, while T12 attenuation <146.2 HU was 61.4% specific and 75.6% sensitive for distinguishing osteoporosis from osteopenia. There were no significant differences between the T10-12 and L1 groups in distinguishing the normal, osteopenia, and osteoporosis groups. Moreover, the diagnostic efficacy among the T10, T11, T12, and L1 vertebral bodies was not statistically significantly different among the three groups.
Opportunistic screening is a valid method for predicting osteopenia or osteoporosis. As a rapid and effective tool, T10-12 vertebral attenuation measures can be incorporated to predict osteoporosis and identify patients who may benefit from further investigations using DXA based on routine clinical chest CT examinations.
由于双能X线吸收法（DXA）是目前最常用的参考标准，大多数先前使用计算机断层扫描（CT）衰减值预测骨质疏松症的研究都选择了腹部CT图像。一些研究调查了与腰椎相比，胸椎是否可独立用于骨质疏松症的识别。
以中心DXA作为参考标准，研究在胸部CT上使用人工智能（AI）测量的胸椎衰减值是否能独立预测骨质疏松症的识别。
这是一项横断面研究。
纳入553名在1天内接受胸部CT和DXA检查的参与者（353名男性和200名女性）。通过AI获得T7-12椎体和L1椎体的衰减值（HU）。比较正常、骨量减少和骨质疏松组的临床基线数据和衰减值的影响。分析衰减与骨密度（BMD）值之间的相关性，并进一步探讨胸椎和第一腰椎椎体衰减值对诊断骨量减少或骨质疏松症的诊断性能。
T7-12和L1椎体的CT衰减值与T值呈正相关（r = 0.58-0.61，P < 0.01）。T12衰减>184.8 HU区分正常骨密度的敏感度为84.1%，特异度为70.6%，而T12衰减<146.2 HU区分骨质疏松症与骨量减少的特异度为61.4%，敏感度为75.6%。在区分正常、骨量减少和骨质疏松组方面，T10-12和L1组之间无显著差异。此外，在三组中，T10、T11、T12和L1椎体的诊断效能在统计学上无显著差异。
机会性筛查是预测骨量减少或骨质疏松症的有效方法。作为一种快速有效的工具，T10-12椎体衰减测量可用于预测骨质疏松症，并根据常规临床胸部CT检查识别可能受益于进一步DXA检查的患者。
