基于三联合重建方案实现全髋关节置换术后假体周围血管的可视化和金属伪影去除的应用研究OA
Application of Multimodal Reconstruction Scheme on Periprosthetic Vasculature Visualization and Metal Artifact Reduction in Total Hip Arthroplasty
目的:探讨能谱 CT单能量成像技术联合去金属伪影(MAR)算法与深度学习重建(DLR)算法对髋关节置换术(THA)后下肢动脉 CT血管成像(CTA)图像质量的改善价值.方法:回顾性纳入 24例髋关节置换术后行下肢动脉双能 CTA检查的患者.对双能 CTA 原始数据进行 4种方案重建:A组(常规混合能量图像)、B组(常规图像+MAR)、C组(常规混合能量图像+DLR)和 D组(单能量+MAR+DLR).比较 4组图像的主观评分(5分制)与客观指标(金属伪影层面股动脉及邻近皮下脂肪的 CT值、标准差(SD),信噪比(SNR)、对比噪声比(CNR)和伪影指数(AI)).采用方差分析和加权 Kappa检验进行统计学比较.结果:4组间除 SNR指标,其余客观指标的差异均具有统计学意义;在 D 组中,随着千电子伏特(keV)升高,伪影指数(AI)持续下降,但血管强化、SNR及 CNR随之降低.60 keV整体评分最高(4.89±0.32),实现血管显示与伪影抑制的最佳平衡.当能级>70 keV后,AI值与血管噪声的变化均趋于平缓.结论:本研究验证能谱 CT单能量成像、MAR算法与 DLR技术的三联方案在改善 THA术后下肢动脉 CTA 图像质量方面的可行性;不同单能量水平在抑制伪影与增强血管对比度之间存在明确权衡.推荐将 60 keV图像作为血管评估的首选能级,以实现主观与客观指标的最佳平衡;并将>70 keV高能级图像作为伪影严重区域的补充评估序列.
Objective:To explore the combination of spectral computed tomography(CT)monochromatic imaging with a metal artifact reduction(MAR)algorithm and deep learning reconstruction(DLR).We also aimed to evaluate the image quality improvements in computed tomography angiography(CTA)of the lower extremity arteries after total hip arthroplasty(THA)using the DLR algorithm.Methods:A total of 24 patients who underwent dual-energy CTA examination of the lower-extremity arteries after hip replacement were retrospectively enrolled.The original dual-energy CTA data were reconstructed using four schemes:group A(conventional mixed-energy images),group B(conventional image+MAR),group C(conventional mixed-energy image+DLR),and group D(single energy+MAR+DLR).The subjective scores(5-point scale)of the four groups of images were compared with the objective parameters CT value,noise(SD),signal-to-noise ratio(SNR),contrast-to-noise ratio(CNR),and artifact index(AI))of the femoral artery and adjacent subcutaneous fat at the metal artifact level.Analysis of variance and the weighted Kappa test were used for statistical comparisons.Results:Except for SNR,the differences in the other objective indices among the four groups were significant.In group D,AI decreased with increasing kiolelectron volt(keV);however,vessel enhancement,SNR,and CNR decreased with increasing keV.The 60 keV images had the highest overall score(4.89±0.32),which achieved the best balance between vessel display and artifact suppression.When the energy level was>70 keV,the changes in the AI value and the vessel noise plateaued.Conclusion:The triple scheme of spectral CT monochromatic energy imaging(60 keV),MAR algorithm,and DLR technology can effectively suppress metal artifacts and improve the definition of vascular structures after THA,which has certain clinical application value.
姚悦;何拓;焦粉;张静凡;刘建新;李晓会
西安交通大学第二附属医院医学影像科,西安 710004西安交通大学第二附属医院医学影像科,西安 710004西安交通大学第二附属医院医学影像科,西安 710004西安交通大学第二附属医院医学影像科,西安 710004西安交通大学医学部基础医学院神经生物学研究所,西安 710061||西安交通大学医学部转化医学研究院神经科学研究所,西安 710061西安交通大学第二附属医院医学影像科,西安 710004
信息技术与安全科学
单能量成像去金属伪影深度学习重建髋关节置换CT血管成像
single-energy imagingmetal artifact removaldeep learning reconstructiontotal hip arthroplastyCT angiography
《CT理论与应用研究》 2026 (6)
1061-1068,8
陕西省"双链"融合国家医学中心医工交叉项目(微剂量CT算法应用验证(2021LL-JB-06)).
评论