ISSN 1004-4140
    CN 11-3017/P

    基于人工智能迭代重建的肺部低剂量CT体模与动物研究

    Low-Dose Pulmonary CT Using Artificial Intelligence Iterative Reconstruction: Phantom and Animal Studies

      摘要
    • 摘要: 目的：探讨借助人工智能迭代重建算法（AIIR），低剂量CT（LDCT）图像能否达到诊断图像质量标准，以用于肺部异常征象的筛查，并确定合适的扫描方案。方法：选择不同CT扫描方案得到体模共28组图像：HRCT组（120/100 kVp，100 mAs，IR/AIIR）和LDCT组（120/100 kVp，50/30/25/20/15/10 mAs，IR/AIIR）；动物共14组图像：对照组（120 kVp，自动管电流，IR/AIIR）和LDCT组（120 kVp，50/30/25/20/15/10 mAs，IR/AIIR）。对体模各组图像进行客观评价，比较各组的图像质量评价参数；对动物各组图像进行主、客观评价，比较各组图像肺组织、主动脉、骨骼肌区域的信噪比，并对总体图像质量、肺部正常结构以及肺部病变区域进行五分制评分。结果：AIIR算法重建图像的TTF50%、NPS peak值、SNR及图像质量评分均优于IR算法。LDCT-AIIR组（120 kVp，25 mAs、30 mAs、50 mAs）与对照组的差异均无统计学意义；LDCT-AIIR组（120 kVp，30 mAs 、 50 mAs）除胸膜下血管外，其余七种主观评价参数的评分与对照组的差异均无统计学意义。LDCT（120 kVp，30 mAs，AIIR）满足诊断图像质量且辐射剂量最低，降低70.09%。结论：LDCT（120 kVp，30 mAs）联合AIIR可以在保持诊断图像质量的情况下，显著减少辐射剂量，为临床肺部低剂量CT扫描提供新方案。

       

      Abstract: Objective: To investigate whether low-dose CT (LDCT) images processed with artificial intelligence iterative reconstruction (AIIR) can achieve a diagnostic image quality comparable to that of high-resolution CT (HRCT) for the detection of pulmonary imaging abnormalities, and to identify a suitable scanning protocol. Methods: Phantom and two adult pigs underwent CT scanning using various protocols. A total of 28 phantom datasets were acquired and divided into high-resolution CT (HRCT) groups (120/100 kVp, 100 mAs, IR/AIIR) and LDCT (120/100 kVp, 50/30/25/20/15/10 mAs, IR/AIIR) groups. Fourteen datasets were collected from the animals, including a control group (120 kVp, IR/AIIR) and an LDCT group (120 kVp, 50/30/25/20/15/10 mAs, IR/AIIR). Phantom images were objectively evaluated in order to compare the image quality parameters. The objective evaluation of animal images focused on the signal-to-noise ratio (SNR) of the lung tissue, aorta, and skeletal muscle regions. Subjective evaluation of animal images focused on the overall image quality, normal lung structures, and pulmonary lesion visualization to determine a suitable low-dose protocol. Results: The AIIR-reconstructed images demonstrated significantly better TTF50%, NPS peak values, SNR, and image quality scores than IR images. For the LDCT-AIIR groups (120 kVp 25 mAs, 30 mAs, and 50 mAs), no statistically significant differences were observed when compared with the control group (P > 0.05). For the LDCT-AIIR groups (120 kVp, 30 mAs, and 50 mAs), no statistically significant differences were observed in the seven subjective image quality parameters (excluding the subpleural vessels) relative to the control group (P > 0.05). The LDCT protocol (120 kVp, 30 mAs, AIIR) achieved diagnostic image quality with the lowest radiation dose, reducing the exposure by 70.09%. Conclusion: LDCT (120 kVp, 30 mAs) combined with AIIR effectively maintained diagnostic image quality while significantly reducing the radiation dose, thus furnishing a clinically viable low-dose protocol for pulmonary computed tomography.

       

      • HTML全文
      • 参考文献(23)
      • 相关文章
      • 施引文献
    • 资源附件(0)

    /

    返回文章
    返回