-
骨质疏松症是一种好发于老年人群、癌症病人和绝经后女性的代谢性骨病,在日常诊疗过程中常被忽视,易导致骨折风险增加、癌症病人预后不良。定量CT(quantitative computed tomography,QCT)是一种基于常规平扫的数据挖掘方法,常作为骨密度(bone mineral density, BMD)测量的参照标准,但目前QCT专用体模和软件尚未完全普及[1-2]。利用高低两组电压瞬切技术实现的双能CT成像(dual-energy CT, DECT)可以提供更多的物质定量信息。DECT虚拟平扫技术去除增强图像中的碘来获取与真实常规平扫(true non-contrast scan, TNC)相当的虚拟平扫(virtual non-contrast scan, VNC)图像[3],虚拟平扫技术由于减少了CT平扫在达到相似的诊断效能同时,还可以降低X线辐射剂量,因此在腹部病变检查中获得良好应用。既往研究以双能X线吸收法(dual-energy X-ray absorptiometry,DXA)为参照标准来验证DECT虚拟平扫用于BMD测量的可行性,但方法繁琐,BMD测量值误差较大。本研究探讨DECT虚拟平扫在无QCT体膜条件下定量测量椎体BMD与评估骨质疏松与骨量减少的可行性及其应用价值。
-
回顾性收集100例于安徽省立医院行腹部DECT常规模式平扫与能谱模式三期增强扫描的肿瘤病人,并随机分为训练集和验证集各50例。纳入标准:(1)有完整的腹部DECT TNC和能谱模式增强扫描图像资料;(2)腹部CT图像包括完整的L1椎体。排除标准:(1)椎体骨折和手术病史;(2)椎体良、恶性肿瘤或恶性肿瘤骨转移者;(3)椎体存在囊性、硬化性病变;(4)严重肝肾功能不全及心功能不全者。本研究所有病人均签署知情同意书,并经我院伦理委员会审核批准。训练集与验证集病人性别、年龄和体质量差异均无统计学意义(P>0.05)(见表 1),具有可比性。
分组 男 女 年龄/岁 体质量/kg 身高/cm 训练集 29 21 57.50±9.70 60.14±13.23 162.19±7.86 验证集 32 18 59.89±11.70 59.98±9.17 164.25±7.01 t 0.38* 0.85 0.07 1.38 P >0.05 >0.05 >0.05 >0.05 *示χ2值 表 1 训练集和验证集病人一般资料比较(x±s)
-
采用GE Discovery CT 750HD(GE Healthcare, Milwaukee, WI, USA)扫描仪。病人取仰卧位,扫描范围自膈肌上缘至耻骨联合下缘。(1)TNC参数:管电压120 kV,自动管电流,噪声指数(NI, noise index) 9,探测器宽度40 mm。螺距1.375∶1,球管转速0.7 s/rot,DFOV 50 cm×50 cm,扫描层厚和层距均为5 mm,重建层厚与层间距均为1.25 mm,扫描结束后采用50% ASIR-V迭代技术获得重建图像。采用异步QCT技术,每日扫描一次体膜进行校准。(2)GSI模式增强扫描参数:采用经外周静脉团注非离子的对比剂碘佛醇(350 mg/mI),流速3.0 mL/s,对比剂用量(1.2 mL/kg)。应用prep smart技术触发扫描,监测点位于T12/L1椎体水平腹主动脉,触发阈值为120 HU,达到阈值5 s后触发扫描,在动脉期结束后29 s行门静脉期扫描,并在门静脉期结束后110 s进行延迟期扫描。管电压80/140 kVp(0.25 ms瞬时切换),管电流375 mA,其他参数及重建方式与常规平扫一致。
-
将常规平扫1.25 mm数据上传至QCT Pro工作站。于L1椎体中心层面放置感兴趣区(region of interesting, ROI),大小10 cm3,避开皮质骨及椎体后中央静脉沟,尽量包括更多的松质骨(见图 1)。所有数据测量均由一名高年资主治医师和一名住院医师独立完成。
-
将TNC和门静脉期1.25 mm数据上传至GE AW4.6工作站,运用GSI Volume Viewer软件获得VNC图像。分别在TNC和VNC图像中于L1椎体中心层面放置ROI(10 cm3),大小和范围尽量与QCT测量保持一致。测量TNC、VNC图像中的L1椎体CT值及其标准差SD(见图 2、3)。尽量避开皮质骨及椎体后中央静脉沟,并包括更多的松质骨。
-
分别测量计算TNC、VNC图像的图像噪声、信噪比(signal noise ratio, SNR),以竖脊肌为参照,计算图像的对比噪声比(contrast noise ratio, CNR)。
分别记录扫描结束后生成的TNC和增强扫描剂量长度乘积(dose length product, DLP),有效辐射剂量(effective dose, ED)计算方法:ED=DLP×k(欧盟委员会推荐成人腹部k值为0.015)。
-
通过logistic多元线性回归分析建立预测椎体BMD值的回归模型,比较该模型测量的BMD与QCT测量的BMD的诊断效能。骨质疏松的诊断采用的标准如下:骨量正常(BMD>120 mg/cm3);骨量减少(BMD为80~120 mg/cm3);骨质疏松(BMD<80 mg/cm3)[4]。
-
采用SPSS 16.0(IBM,Armonk,NY)和MedCalc 15.2.2 (Ostend,Belgium)统计分析软件。采用ICC评价2名医师测量数据的一致性,ICC>0.75位一致性良好。若一致性良好选择高年资主治医师测量数据进行统计学分析,采用t检验和χ2检验。在训练集中采用VNC椎体CT值与椎体BMD值间的相关性采用Spearman相关分析。通过logistic回归分析建立预测椎体BMD值的回归模型。在验证集中采用Bland-Altman法分析验证集椎体BMD回归与BMD实际间数据的一致性,并通过ROC曲线分析VNC椎体CT值对骨质疏松和骨量减少的诊断价值。
-
100例病人TNC、VNC图像的竖脊肌CT值、椎体和竖脊肌的噪声SD差异无统计学意义(P>0.05)(见表 2)。TNC、VNC图像的SNR和CNR差异均有统计学意义(P < 0.01)。TNC的ED为(9.44±2.15)mSv,DLP为(629.23±143.66)mGy·cm;VNC的DLP为(531.14±100.74)mGy·cm;总DLP为(2 867.96±127.69)mGy·cm。在减去常规平扫后,行增强CT检查的病人总辐射剂量减少了21.94%。
椎体CT值 竖脊肌CT值 CT值 SD椎体 SD竖脊肌 SNR CNR TNC 133.57±48.4 48.26±6.08 17.24±3.76 10.32±1.30 12.34±4.93 7.49±4.66 VNC 71.24±26.38 52.46±5.06 19.83±4.58 11.08±1.50 5.73±2.48 1.87±2.28 t 8.00 4.60 11.03 3.31 10.38 11.27 P < 0.05 >0.05 >0.05 >0.05 < 0.05 < 0.05 表 2 TNC、VNC图像质量比较(x±s)
-
训练集VNC的L1椎体CT值较TNC低46.90%,两者差异有统计学意义(P < 0.01)。VNC的L1椎体CT值与QCT测量的BMD呈显著正相关(r=0.974, P < 0.01)(见图 4)。
多元线性回归分析结果显示病人年龄、身高、VNC椎体CT值与QCT测量的BMD值相关,病人性别、体质量与其BMD值间不存在线性关系(P>0.05)。经拟合,得到的BMD多元线性回归方程为:BMD=102.375+1.349×CT值-0.321×年龄-0.433×身高(R2=0.979)(见表 3)。
项目 B SE B′ t P 常数 102.375 34.718 — 2.95 < 0.05 VNC CT值 1.349 0.440 0.938 30.98 < 0.01 年龄 -0.321 0.110 -0.095 2.92 < 0.01 性别 3.339 2.864 0.035 1.17 >0.05 体质量 0.055 0.120 0.011 0.45 >0.05 身高 -0.433 0.205 -0.071 2.11 < 0.05 表 3 VNC椎体CT值与QCT测量的BMD间的多元线性回归分析
-
验证集的BMD回归与BMD实际差异无统计学意义(P>0.05)。通Bland-Altman散点图显示除个别异常值以外,验证集BMD回归与BMD实际之间的差值均匀的分布95%的一致性界限范围内(95%CI:-14.201~20.523, P>0.05)(见图 5)。
-
VNC椎体CT值诊断骨质疏松、骨量减少的曲线下面积分别为AUC骨松=0.976(95%CI:0.922~0.997)、AUC骨少=0.964(95%CI:0.913~0.991),并根据约登指数确定骨质疏松、骨量减少最佳诊断界值分别为54.35 HU(敏感性:87.53%;特异性:92.31%)、64.51 HU(敏感性:84.78%;特异性:89.42%)(见图 6、7)。
双能CT虚拟平扫定量评估骨质疏松的可行性研究
Study on the feasibility of quantitative evaluation of osteoporosis by dual-energy CT virtual non-contrast scan
-
摘要:
目的以QCT测量的骨密度(BMD)为金标准,探讨双能CT(DECT)虚拟平扫在无体模的状态下定量评估BMD与骨质疏松程度的可行性。 方法回顾性收集100例行腹部混合能量常规平扫(TNC)与能谱模式(GSI)三期增强扫描的病人,并随机分为训练集和验证集各50例。使用QCT Pro软件测量L1椎体BMD值。在训练集中比较常规模式平扫图像和GSI模式增强门静脉期虚拟平扫图像中L1椎体CT值差异,并确定虚拟平扫椎体CT值与BMD值的相关性,通过多元回归分析建立虚拟平扫椎体CT值预测其BMD值的回归模型。验证集用来评估该方程诊断骨质疏松的准确性以及虚拟平扫椎体CT值对骨质疏松的诊断效能。 结果虚拟平扫的L1椎体CT值低于常规平扫(P < 0.01);在减去常规平扫后图像的总辐射剂量减少了21.94%;虚拟平扫CT值预测椎体BMD值的回归模型:BMD=102.375+1.349×CT值-0.321×年龄-0.433×身高(R2=0.979);经ROC曲线分析发现基于门静脉期的虚拟平扫椎体CT值诊断骨质疏松和骨量减少的截断值分别为54.35 HU和64.51 HU。 结论在不依赖体模的情况下,DECT虚拟平扫的椎体CT值能够可靠测量BMD,通过建立的预测BMD模型,可定量诊断骨质疏松和骨量减少,是QCT技术的有益补充。 Abstract:ObjectiveTo investigate the feasibility of dual-energy CT(DECT) virtual non-contrast scan in the quantitative evaluation of bone mineral density(BMD) measured by QCT as the gold standard in the absence of a body mode. MethodsA retrospective study was performed on 100 patients with abdominal true non-contrast scan(TNC) and GSI three-phases enhanced scan, and the patients were randomly divided into the training set and validation set(50 cases in each set). The BMD value of L1 vertebral body in two sets was measured using QCT Pro software. In the training set, the difference of L1 vertebral CT value between conventional non-contrast scan image and GSI enhanced virtual non-contrast scan image in portal vein phase was compared, and the correlation between virtual non-contrast scan vertebral CT value and BMD value was determined. The regression model of virtual non-contrast scan vertebral CT value to predict BMD value was established through multiple regression analysis. The validation set was used to evaluate the accuracy of this equation in the diagnosis of osteoporosis and effectiveness of virtual non-contrast scan vertebral CT value in the diagnosis of osteoporosis. ResultsThe CT value of L1 vertebral body of VNC was lower than that of TNC(P < 0.01). After subtracting the TNC, the total radiation dose of image was reduced by 21.94%, the multiple linear regression model of CT value of VNC to predict BMD of L1 vertebra was as follows: BMD=102.375+1.349×CT value-0.321×age-0.433×height(R2=0.979). The ROC curve analysis showed that the cut-off value of virtual non-contrast scan vertebral body CT value based on portal vein phase in the diagnosis of osteoporosis and osteopenia were 54.35 HU and 64.51 HU, respectively. ConclusionsIn the case of independent body model, the DECT virtual non-contrast scan vertebral CT value can reliably measure the BMD, and the established predictive BMD model can quantitatively diagnose the osteoporosis and osteopenia, which is a beneficial supplement to QCT technology. -
Key words:
- virtual non-contrast scan /
- bone mineral density /
- osteoporosis /
- dual-energy CT
-
表 1 训练集和验证集病人一般资料比较(x±s)
分组 男 女 年龄/岁 体质量/kg 身高/cm 训练集 29 21 57.50±9.70 60.14±13.23 162.19±7.86 验证集 32 18 59.89±11.70 59.98±9.17 164.25±7.01 t 0.38* 0.85 0.07 1.38 P >0.05 >0.05 >0.05 >0.05 *示χ2值 表 2 TNC、VNC图像质量比较(x±s)
椎体CT值 竖脊肌CT值 CT值 SD椎体 SD竖脊肌 SNR CNR TNC 133.57±48.4 48.26±6.08 17.24±3.76 10.32±1.30 12.34±4.93 7.49±4.66 VNC 71.24±26.38 52.46±5.06 19.83±4.58 11.08±1.50 5.73±2.48 1.87±2.28 t 8.00 4.60 11.03 3.31 10.38 11.27 P < 0.05 >0.05 >0.05 >0.05 < 0.05 < 0.05 表 3 VNC椎体CT值与QCT测量的BMD间的多元线性回归分析
项目 B SE B′ t P 常数 102.375 34.718 — 2.95 < 0.05 VNC CT值 1.349 0.440 0.938 30.98 < 0.01 年龄 -0.321 0.110 -0.095 2.92 < 0.01 性别 3.339 2.864 0.035 1.17 >0.05 体质量 0.055 0.120 0.011 0.45 >0.05 身高 -0.433 0.205 -0.071 2.11 < 0.05 -
[1] CHENG X, LI K, ZHANG Y, et al. The accurate relationship between spine bone density and bone marrow in humans [J]. Bone, 2020, 134: 115312. doi: 10.1016/j.bone.2020.115312 [2] XU XM, LI N, LI K, et al. Discordance in diagnosis of osteoporosis by quantitative computed tomography and dual-energy X-ray absorptiometry in Chinese elderly men[J]. J Orthop Translat, 2018, 18: 59. [3] SAUTER AP, MUENZEL D, DANGELMAIER J, et al. Dual-layer spectral computed tomography: Virtual non-contrast in comparison to true non-contrast images[J]. Eur J Radiol, 2018, 104: 108. doi: 10.1016/j.ejrad.2018.05.007 [4] 中华医学会放射学分会骨关节学组, 中国医师协会放射医师分会肌骨学组, 中华医学会骨科学分会骨质疏松学组, 等. 骨质疏松的影像学与骨密度诊断专家共识[J]. 中华放射学杂志, 2020, 54(8): 745. doi: 10.3760/cma.j.cn112149-20200331-00485 [5] 李长洲, 庞炎旭, 于利, 等. 骨密度在髋部骨质疏松性骨折风险评估中的价值[J]. 中国骨质疏松杂志, 2020, 26(7): 1023. [6] MILISIC L, VEGAR-ZUBOVIC S, VALJEVAC A, et al. Bone mineral density assessment by DXA Vs. QCT in postmenopausal females with central obesity[J]. Curr Aging Sci, 2020, 13(2): 153. doi: 10.2174/1874609812666190912155525 [7] JAVADI S, ELSHERIF S, BHOSALE P, et al. Quantitative attenuation accuracy of virtual non-enhanced imaging compared to that of true non-enhanced imaging on dual-source dual-energy CT[J]. Abdom Radiol(NY), 2020, 45(4): 1100. doi: 10.1007/s00261-020-02415-8 [8] HAJI-MOMENIAN S, PARKINSON W, KHATI N, et al. Single-energy non-contrast hepatic steatosis criteria applied to virtual non-contrast images: is it still highly specific and positively predictive?[J]. Clin Radiol, 2018, 73(6): 594. e7. doi: 10.1016/j.crad.2018.01.018 [9] PAPADAKIS AE, KARANTANAS AH, PAPADOKOSTAKIS G, et al. Can abdominal multi-detector CT diagnose spinal osteoporosis?[J]. Eur Radiol, 2009, 19(1): 172. doi: 10.1007/s00330-008-1099-2 [10] SHAO HD, LI GW, LIU Y, et al. Contributions of fat mass and fat distribution to hip bone strength in healthy postmenopausal Chinese women[J]. J Bone Miner Metab, 2015, 33(5): 507. doi: 10.1007/s00774-014-0613-7 [11] SHINODA Y, OGATA N, KAWAGUCHI H. Regulation of bone formation by adiponectin through autocrine/paracrine and endocrine pathways[J]. Nihon Rinsho, 2007, 65(9): 90. [12] ZHANG X, HUA T, ZHU J, et al. Body compositions differently contribute to BMD in different age and gender: a pilot study by QCT[J]. Arch Osteoporos, 2019, 14(1): 31. doi: 10.1007/s11657-019-0574-5 [13] MEI K, SCHWAIGER BJ, KOPP FK, et al. Bone mineral density measurements in vertebral specimens and phantoms using dual-layer spectral computed tomography [J]. Sci Rep, 2017, 7(1): 17519. doi: 10.1038/s41598-017-17855-4 [14] DING Y, RICHTER A, STILLER W, et al. Association between true non-contrast and virtual non-contrast vertebral bone CT attenuation values determined using dual-layer spectral detector CT [J]. Eur J Radiol, 2019, 121: 108740. doi: 10.1016/j.ejrad.2019.108740 [15] 武卫杰, 岳松伟, 王会霞, 等. 能谱CT虚拟平扫评估肺纯磨玻璃结节的可行性研究[J]. 放射学实践, 2020, 35(3): 346. [16] 杨泠, 瞿姣, 袁欣, 等. 双能CT迭代重建技术联合虚拟平扫对胃壁图像质量及辐射剂量的影响[J]. 中国临床医学影像杂志, 2019, 30(1): 61. [17] 杨琰昭, 严福华, 韩群, 等. 双层光谱探测器CT腹部虚拟平扫代替常规平扫的可行性研究[J]. 中华放射学杂志, 2019, 53(1): 33. doi: 10.3760/cma.j.issn.1005-1201.2019.01.008