| 胡永志,王彬彬,朱亚新,武鹏飞,曹达,唐玉霞,王传兵,崔维顶,王守巨.深度学习全模型迭代算法用于髋关节置换术后腹盆腔CT检查[J].中国医学影像技术,2025,41(4):553~556 |
| 深度学习全模型迭代算法用于髋关节置换术后腹盆腔CT检查 |
| Artificial intelligence iterative reconstruction for abdominal and pelvic CT examination after total hip arthroplasty |
| 投稿时间：2024-10-21 修订日期：2025-04-10 |
| DOI：10.13929/j.issn.1003-3289.2025.04.010 |
| 中文关键词: 体层摄影术,X线计算机 关节成形术,置换 人工智能 |
| 英文关键词:tomography,X-ray computed arthroplasty, replacement artificial intelligence |
| 基金项目:江苏省人民医院高层次人才培育计划(CZ0121002010039)。 |
|
| 摘要点击次数: 536 |
| 全文下载次数: 103 |
| 中文摘要: |
| 目的 观察深度学习全模型迭代算法(AIIR)用于全髋关节置换术(THA)后腹盆腔CT的价值。方法 回顾性收集接受腹盆腔CT检查的64例THA后患者,其中31例接受常规CT扫描、33例接受双能CT扫描。对常规CT图像分别采用AIIR及混合迭代重建(HIR)算法获得AIIR及HIR图像,对双能CT重建70~140 keV虚拟单能量图像(VMI)(间隔5 keV),选择综合图像质量最佳图像进行分析;对比3种图像质量主、客观评价结果。结果 AIIR图及VMI的伪影、骨骼、诊断信心及显示盆腔器官和血管主观评分均高于HIR图(P均＜0.001),而AIIR图与VMI上述主观评分差异均无统计学意义(P均＞0.017)。AIIR、HIR图及VMI中的高密度伪影分数及骨骼伪影分数两两比较差异均有统计学意义(P均＜0.001)。AIIR图与VMI之间低密度伪影分数及高密度噪声分数差异均无统计学意义(P均＞0.017),二者的上述客观评价指标与HIR图差异均有统计学意义(P均＜0.017)。AIIR图的低密度噪声分数低于HIR图(P＜0.017),而AIIR及HIR图的低密度噪声分数与VMI差异均无统计学意义(P均＞0.017)。AIIR及HIR图的骨骼噪声分数均高于VMI(P均＜0.017),而二者之间骨骼噪声分数差异无统计学意义(P＞0.017)。结论 AIIR可减少THA后腹盆腔CT伪影、降低图像噪声并改善图像质量。 |
| 英文摘要: |
| Objective To observe the value of artificial intelligence iterative reconstruction (AIIR) for abdominal and pelvic CT examination after total hip arthroplasty (THA). Methods Totally 64 patients after THA who underwent abdominal and pelvic CT examinations were retrospectively collected, including 31 patients received routine CT scanning and 33 patients received dual-energy CT scanning. AIIR and hybrid iterative reconstruction (HIR) algorithms were used to obtain AIIR and HIR images based on conventional CT images, respectively, while 70—140 keV (interval of 5 keV) virtual monoenergetic images (VMI) were reconstructed based on dual-energy CT images. VMI with the best comprehensive imaging qualities were selected for analysis. Subjective scores and objective evaluation results of imaging quality were compared among different kinds of images. Results The subjective scores of artifacts, bones, diagnostic confidence, as well as displaying of pelvic organs and blood vessels on both AIIR images and VMI were all higher than those of HIR images (all P＜0.001), while no significant difference was found between AIIR images and VMI (all P＞0.017). Pairwise comparison of high-density artifact fraction and skeletal artifact fraction on AIIR, HIR images and VMI showed significant differences (all P＜0.001). No significant difference of low density artifact fraction nor high density noise fraction was detected between AIIR image and VMI (both P＞0.017), and the objective evaluation results were different from those of HIR images (both P＜0.017). The low density noise fraction of AIIR images was lower than that of HIR images (P＜0.017), while no significant difference was found between AIIR or HIR images and VMI (both P＞0.017). The bone noise fraction of AIIR and HIR images were both higher than that of VMI (both P＜0.017), while no significant difference was found between these two kinds of images (P＞0.017). Conclusion AIIR could reduce artifacts and image noise of abdominal and pelvic CT examination after THA and improve imaging quality. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
