降低腰椎内固定术后金属伪影:迭代去金属伪影算法联合虚拟单能成像最佳能级的研究OA
Reduction of Metal Artifacts after Lumbar Internal Fixation:A Study Combining an Iterative Metal Artifact Reduction Algorithm with Optimal Kiloelectron-volt-level Virtual Monoenergetic Imaging
目的:评估迭代去金属伪影(iMAR)算法联合不同能级的虚拟单能成像(VMI)技术去金属伪影的效果和图像质量.方法:前瞻性筛选腰椎内固定术后患者,分别不采用和采用 iMAR算法进行 VMI图像重建,重建能级包括 70、100、120、140、160和 190 keV.选取螺钉长轴显示最佳的横断面图像(伪影层)和无明显伪影横断面图像(背景层),在椎体、椎管、竖脊肌和皮下脂肪勾画感兴趣区(ROI),记录其 CT值和标准差(SD)值.计算伪影层与背景层的 CT差异值(ΔCT)、平均背景 SD 值、平均伪影 SD 值和伪影指数(AI).对图像的评价金属伪影去除效果和对于诊断的影响进行主观评分.比较主客观评价结果.结果:研究共纳入 65名患者,年龄中位数(四分位数)为 65.00(56.50,72.00)岁,男性 32名(47.76%),女性 33名(50.76%).扫描容积 CT剂量指数为 16.55(14.79,19.14)mGy,剂量长度乘积为 458.80(410.05,542.75)mGy·cm.70~160 keV 同一能级的图像中,与不采用 iMAR算法的图像相比,采用 iMAR算法的各个部位的 ΔCT值、平均伪影 SD 值和各个部位的 AI值均显著更低.与 70 keV图像相比,100~190 keV 图像各个解剖结构的 ΔCT值、背景及伪影 SD值和各个部位的 AI值均显著更低.主观评价显示,同一能级的 VMI图像中,采用 iMAR算法能明显改善图像的金属伪影,较高能级的VMI图像可以在一定程度上改善图像的金属伪影;采用iMAR算法联合较高能级的VMI图像可以更有效地去除伪影,其中 120~140 keV图像伪影均较低.结论:本研究推荐 iMAR算法联合 140 keV图像用于腰椎固定术后评估 CT图像的重建,以获得更为理想的图像质量,为临床诊断提供更丰富的诊断信息.
Objective:To evaluate image quality and the effect of metal artifact reduction with and without an iterative metal artifact reduction(iMAR)algorithm combined with virtual monoenergetic imaging(VMI)at different energy levels.Methods:Patients who had undergone internal fixation of the lumbar spine were prospectively enrolled.The VMI images were reconstructed both with and without the iMAR algorithm at energy levels of 70,100,120,140,160,and 190 keV.Axial images with an optimal visualization of the screw along its long axis(artifact layer)and those without obvious artifacts(background layer)were selected.Regions of interest(ROIs)were delineated in the vertebral body,spinal canal,erector spinae,and subcutaneous fat,and CT values and standard deviations(SD)were recorded.The difference in CT values(ΔCT)between the artifact and background layers,mean background SD,mean artifact SD,and artifact index(AI)were calculated.Subjective scoring was performed to assess the metal artifact reduction effect and its impact on diagnosis.The objective and subjective evaluation results were compared.Results:A total of 65 patients with a median age(interquartile range)of 65.00(56.50,72.00)years were included,including 32 males(47.76%)and 33 females(50.76%).The volume CT dose index was 16.55(14.79,19.14)mGy,and the dose–length product was 458.80(410.05,542.75)mGy·cm.For VMI images at the same energy level from 70 to 160 keV,the ΔCT,mean artifact SD,and AI values of all the anatomical sites in images reconstructed with the iMAR algorithm were significantly lower than those without the iMAR.Compared with images at 70 keV,the ΔCT,background and artifact SD,and AI values of all the anatomical structures in images at 100 to 190 keV were significantly lower.Subjective evaluations indicated that,at the same VMI energy level,the iMAR algorithm significantly reduced metal artifacts.VMI images acquired at higher energy levels reduced metal artifacts to a certain extent.The combination of the iMAR algorithm and VMI at higher energy levels achieved more effective artifact reduction,with the lowest artifacts observed at 120 to 140 keV.Conclusion:This study recommends VMI at 140 keV combined with the iMAR algorithm for CT image reconstruction in the postoperative evaluation of lumbar spine fixation,as it yields improved image quality and provides enhanced diagnostic information for clinical diagnosis.
王思连;张欢;姚伟武;钟京谕;陈黎伟;黄普珍;金汉樯;星月;胡扬帆;许芷菡;宋阳;褚敬申
上海交通大学医学院附属同仁医院影像科,上海 200336||上海市柔性医疗机器人重点实验室,上海 200336上海交通大学医学院附属瑞金医院放射科,上海 200050上海交通大学医学院附属同仁医院影像科,上海 200336||上海市柔性医疗机器人重点实验室,上海 200336上海交通大学医学院附属同仁医院影像科,上海 200336||上海市柔性医疗机器人重点实验室,上海 200336上海交通大学医学院附属同仁医院影像科,上海 200336||上海市柔性医疗机器人重点实验室,上海 200336上海交通大学医学院附属同仁医院影像科,上海 200336||上海市柔性医疗机器人重点实验室,上海 200336上海交通大学医学院附属仁济医院骨科,上海 200127上海交通大学医学院附属同仁医院影像科,上海 200336||上海市柔性医疗机器人重点实验室,上海 200336上海交通大学医学院附属同仁医院影像科,上海 200336||上海市柔性医疗机器人重点实验室,上海 200336西门子医疗诊断影像CT科研合作部,上海 200126西门子医疗磁共振科研市场部,上海 200126上海交通大学医学院附属瑞金医院科技发展处《诊断学理论与实践》编辑部,上海 200050
医药卫生
腰椎金属伪影迭代算法虚拟单能成像图像质量
lumbar vertebraemetal artifactiterative algorithmvirtual monoenergetic imagingimage quality
《CT理论与应用研究》 2026 (6)
1029-1038,10
国家自然科学基金项目(基于CT增强影像组学联合代谢组学构建急性胰腺炎重症化智能早期预警模型研究(82302183)基于注意力−可解释图文信息挖掘创建膝骨关节炎交互式诊断报告系统(82471935)厌氧消化链球菌重塑局部进展期直肠癌免疫微环境及MR预测模型研究(82271934))脊柱微创诊疗关键技术与材料实验室开放基金项目(基于人工智能的脊柱退行性病变CT影像研究(2024JZWC-ZDA04)高级弥散模型DXI平台稳健性及其在腰椎间盘退变定量评价及突出定位诊断中的应用研究(2024JZWC-YBA07)).
评论