不同重建算法与辐射剂量对定量CT骨密度测量的影响:体模研究  

Investigating the effect of different reconstruction algorithms and radiation doses on quantitative CT bone mineral density measurement:a phantom study

作　　者：童小雨 刘义军[1] 方鑫[1] 李贝贝 王旭[1] 陈安良[1] TONG Xiao-yu;LIU Yi-jun;FANG Xin(Department of Radiology,the First Affiliated Hospital of Dalian Medical University,Liaoning 116011,China)

机构地区：[1]大连医科大学附属第一医院放射科,辽宁大连116011

出　　处：《放射学实践》2023年第8期1055-1059,共5页Radiologic Practice

摘　　要：目的:探讨不同辐射剂量、不同等级的滤波反投影算法(FBP)、迭代重建算法(Karl)和深度学习重建算法(DL)对骨密度测量值准确性的影响。方法:使用联影uCT760 CT机,分别采用剂量调制等级3(常规辐射剂量组,ND)和剂量调制等级1(低辐射剂量组,LD)对拟人化欧洲脊柱体模(ESP)进行10次重复扫描。扫描结束后对每次扫描的图像分别采用FBP、Karl(5、7、9级)和DL(1~4级)算法进行图像重建,共获得16组重建图像。在QCT Pro工作站上测量各组图像上L_(1)~L_(3)椎体的骨密度(BMD)值,并计算其与各椎体体模标定值的相对误差(RE);在L_(1)、L_(2)和L_(3)椎体内放置感兴趣区,测量其CT值及SD值,并计算对比噪声比(CNR)。采用独立样本t检验分析分别比较16组图像中BMD测量值与真实值的差异,采用配对样本t检验分析ND组与LD组之间BMD测量值和辐射剂量的差异,采用单因素方差分析比较相同辐射剂量不同重建算法和等级的8组图像之间BMD测量值和CNR的差异。结果:LD组的毫安秒(mAs)、CTDIvol和DLP相较于ND组分别降低了78.41%、78.25%和78.28%,差异均有统计学意义(t=158.692~1699.766,P均<0.001)。不同辐射剂量相同重建算法下,L_(1)、L_(2)和L_(3)的BMD值的差异均无统计学意义(t=-2.006~1.472,P均>0.05);相同辐射剂量下8组重建图像上所测得的L_(1)~L_(3)椎体的BMD值的总体差异均无统计学意义(F=0.485~1.954,P均>0.05);各组图像所测得的BMD值与ESP体模标定值之间的差异均有统计学意义(P均<0.05),RE范围为2.17%~7.44%;相同辐射剂量组内8组重建图像上测得的各椎体CNR的差异均有统计学意义(F=47.883~153.983,P均<0.001)。结论:不同重建算法及相同算法不同等级在不同辐射剂量下对骨密度测量的准确性均无显著影响;选择合适的重建算法及等级可在保证图像质量的前提下有效降低患者的辐射剂量。Objective:To investigate the effect of different levels of filtered back-projection algorithm(FBP),iterative reconstruction algorithm(Karl)and deep learning reconstruction algorithm(DL)on the accuracy of bone mineral density assessment at different radiation doses.Methods:Using the United Image uCT760 CT scanner,a phantom was scanned at different radiation doses by changing the level of dose modulation technique.The anthropomorphic European Spine Phantom(ESP)was scanned 10 times with dose modulation level 3(conventional radiation dose group,ND)and dose mo-dulation level 1(low radiation dose group,LD),respectively.After scanning,the images of each set were reconstructed using FBP,Karl(level 5,7,9)and DL(level 1~4)algorithms,respectively,and a total of 16 groups of reconstructed images were obtained.The QCT Pro workstation was used to mea-sure the bone mineral density(BMD)of L_(1)~L_(3) vertebral bodies in each set of images and the relative errors(RE)were calculated.Regions of interest were placed in the L_(1),L_(2) and L_(3) vertebrae,and the CT and its SD values were measured to calculate the contrast-to-noise ratio(CNR).The two-sample t-test was used to compare the BMD measurements on 16 groups of images with the true BMD values.The paired t-test was used to analyze the differences in BMD measurements and radiation doses between ND group and LD group.One-way ANOVA was used to compare the differences in BMD measurements and CNR among the 8 groups of images with different reconstruction algorithms and levels at the same radiation dose.Results:The mean of mAs,CTDIvol and DLP in the LD group were reduced by approximately 78.41%,78.25%and 78.28%compared with the ND group,all with statistically significant differences(t=158.692~1699.766,all P<0.001).There were no statistically significant diffe-rences in BMD measurement of L_(1),L_(2) and L_(3) by ESP at different radiation doses and with the same reconstruction algorithm and level(t=-2.006~1.472,all P>0.05).At the same radiation dose,there was no significant differe

关 键 词：图像重建 深度学习 迭代重建 骨密度 辐射剂量 体层摄影术 X线计算机 

分 类 号：R814.42[医药卫生—影像医学与核医学] R681.5[医药卫生—放射医学]