新一代基于模型的迭代重建中肺特异性设置r在提高胸部CT图像质量中的应用价值

摘要

目的:比较自适应统计迭代重建(ASIR)、常规基于模型的迭代重建(MBIRc)、新一代基于模型的迭代重建(MBIRn)中肺特异性设置(MBIRRP20和MBIRNR40)对常规剂量胸部CT平扫图像质量的影响,探讨MBIRn肺特异性设置在提高胸部CT图像质量中的应用价值.方法:随机抽取我院使用能谱CT(Discovery CT750 HD)行胸部CT平扫的30例受检者.分别采用标准算法与肺算法ASIR40、MBIRc和MBIRn中肺特异性设置(MBIRRP20和MBIRNR40)重建层厚0.625 mm的图像并进行对比分析.在标准算法ASIR、MBIRc和MBIRNR40重建图像上测量胸廓入口层面、气管隆突下层面和所及上腹肝门层面背部肌肉、皮下脂肪相同部位ROI的CT值与标准差值(SD),SD代表噪声,SNR由CT值除以SD计算出,采用单因素方差分析比较各重建算法SD和SNR.由2名放射科医师以ASIR为基础,使用半定量目测评分法(-3~3)在肺窗ASIR、MBIRc、MBIRRP20和纵隔窗ASIR、MBIRc、MBIRNR40双盲法进行噪声和细节结构、病变边缘清晰度评分,采用Wilcoxon符号等级检验主观评分差异性.结果:MBIRn中肺特异性设置MBIRNR40重建图像肌肉噪声、脂肪噪声,均低于MBIRc重建和ASIR重建(均P<0.05).MBIRn中肺特异性设置MBIRNR40重建图像肌肉SNR、脂肪SNR、绝对值大于MBIRc重建和ASIR(P<0.05).MBIRn重建的主观图像噪声最低,优于MBIRc,MBIRc优于ASIR(P<0.05);MBIRn更清晰显示肺、纵隔、上腹部细节结构和病变边缘特征,优于MBIRc,MBIRc优于ASIR(P<0.05).结论:在常规剂量胸部CT平扫条件下,与MBIRc、ASIR相比,肺特异性设置MBIRn中的MBIRRP20和MBIRNR40可显著降低胸部CT平扫图像的噪声、提高SNR,可更清晰显示扫描范围内的细节结构和病变边缘特征,为进一步降低胸部CT辐射剂量提供可能.%Objective:To assess image quality of adaptive statistical iterative reconstruction (ASIR),conventional model-based iterative reconstruction (MBIRc) and a new lung-specific setting (MBIRRP20 and MBIRNR40) from new version of model-based it-erative reconstruction in image quality of the routine dose chest CT. Methods:This retrospective clinical study randomly includ-ed 30 patients who underwent an enhanced chest CT in our hospital. Raw data sets were reconstructed by using MBIRc, MBIRn and ASIR40 (40% ASIR and FBP mixed) in reconstructed slice thickness of 0.625 mm. Objective image noise (standard deviation,SD) and SNR were measured by placing ROI on the same part from the back muscle and subcutaneous fat which located at the level of thoracic entry,trachea carina and hepatic portal. Two radiologists used a semiquantitative 7-point scale (-3 to +3) to rate subjective image quality of lung,mediastinal and upper abdomen structures comparing both MBIRc and MBIRn images with ASIR images,and comparing MBIRc and MBIRn images simultaneously. Compared the results between ASIR,MBIRc and MBIRn by using one-way ANOVA and Wilcoxon signed-rank tests. Results:The mean image noise for MBIRNR40 from MBIRn were (7.80±1.43) HU in muscle and (8.51±2.45) HU in fat,significantly lower than MBIRc [(11.97± 2.37) HU and (12.16±2.68) HU],and ASIR [(24.55±4.14) HU and (19.20±4.11HU)](P<0.05). The mean SNR for MBIRNR40 were (6.79±1.91) HU in muscle and (-13.96±3.93) HU in fat,significantly higher than MBIRc [(4.66±1.41) HU and (-9.53± 2.56) HU],and ASIR [(2.05±0.47) HU and (-6.80±2.75) HU] (P<0.05). The subjective image noise score for MBIRn was significantly better than with MBIRc and ASIR. The sharpness of details of the structures with MBIRn were significantly better than with MBIRc and ASIR (P<0.05),MBIRc were significantly better than ASIR (P<0.05). Conclusions:MBIRRP20 and MBIRNR40 from MBIRn with lung-specific setting have the ability to reduce image noise and improve the image quality compared with ASIR and MBIRc in routine dose chest CT. In other words,this MBIRn with lung-specific setting can provide more potential to further reduce radiation dose in the chest CT.