新一代基于模型的迭代重建在提高上腹部CT图像质量中的应用价值

Value of New Model-Based Iterative Reconstruction for Improving Image Quality in Routine Dose Upper Abdominal CT: A Phantom and Patient Study

目的比较滤波反投影(Filtered Back Projection,FBP)、常规基于模型的迭代重建(Model-Based Iterative Reconstruction,MBIRc)、新一代基于模型的迭代重建(MBIRn)中优化噪声设置的MBIR_(NR40)对腹部CT图像质量的影响,探讨MBIR_(NR40)在提高上腹部CT图像中的应用价值。方法使用能谱CT(Discovery CT750HD)扫描静止状态下水模以客观比较0.625 mm层厚FBP、MBIRc和MBIR_(NR40)各设置的密度分辨率。回顾性随机抽取我院行上腹部增强扫描180 s延迟期的60例受检者。分别采用标准算法FBP、MBIRc和MBIR_(NR40)重建层厚0.625 mm图像后进行对比。在相同位置皮下脂肪、背部肌肉、肝脾实质放置感兴趣区测量CT值与标准差值(Standard Deviation,SD),计算以皮下脂肪为背景肝脾实质对比噪声比(Contrast to Noise Ratio,CNR),单因素方差分析比较各重建算法SD和CNR。采用Wilcoxon检验主观评分差异性。结果水模结果提示MBIR_(NR40)密度分辨率最高,FBP重建密度分辨率最差。MBIR_(NR40)重建图像噪声低于MBIRc重建和FBP重建(P<0.05)。MBIR重建图像肝脾CNR大于MBIRc重建和FBP(P<0.05)。MBIR_(NR40)重建的主观图像噪声最小,FBP噪声最大,差异有统计学意义(P<0.05);MBIR_(NR40)重建更清晰显示细节结构和病变边缘特征,优于MBIRc,MBIRc优于FBP(P<0.05)。结论与MBIRc、FBP重建相比,MBIR_(NR40)可显著降低上腹部CT图像的噪声、提高信噪比,可更清晰显示细节结构和病变边缘特征,为降低上腹部CT辐射剂量提供潜能。

Objective To compare the objective and subjective image quality of filtered back projection(FBP),conventional modelbased iterative reconstruction(MBIRc)and the low-contrast resolution preference algorithm(MBIRNR40)from new version of model-based iterative reconstruction(MBIRn)at routine dose multidetector computed tomographic(CT)examinations of the upper abdomen.Methods This institutional reviewed board-approved study including 60 consecutive patients,who underwent routine dose CT examinations of the upper abdomen.Raw data from patient sets were reconstructed by using FPB,MBIRc and MBIRNR40 with the optimal setting.Measurements of objective noise and CT numbers of the subcutaneous fat,back muscles,liver and spleen were compared with analysis of variance by using one-way ANOVA test.The image quality was assessed by Wilcoxon.Results The phantom study determined MBIRNR40 with an improved low-contrast detectability compared with MBIRc or FBP.The mean image noise(in HU)for MBIRNR40 were significantly lower than those of MBIRc and FBP(all P<0.05).The mean SNR values for MBIRNR40 were significantly higher than those of MBIRc and FBP(all P<0.05).Subjective ratings showed higher image quality for MBIRNR40,with lower noise,better resolution for details in abdominal structures,and finer contours of lesions(P<0.05),while FBP performed inferiorly compared with MBIRc(P<0.05).Conclusion The MBIRNR40 from MBIRn algorithm significantly improved the objective and subjective image quality in routine dose upper abdominal CT scans compared with MBIRc and FBP algorithms,which can be used to reduce radiation dose.