新一代基于模型的迭代重建中肺特异性设置在亚mSv胸部CT检查中的应用

发布时间：2018-03-15 18:29

本文选题：基于模型的迭代重建　切入点：自适应统计迭代重建　出处：《中国医学影像技术》2017年10期 　论文类型：期刊论文

【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)重建算法对亚mSv胸部CT图像质量的影响。方法收集接受两次胸部CT平扫的受检者30例。初检采用常规剂量(噪声指数=14)ASIR重建。复查采用低辐射剂量方案(噪声指数=28),分别采用标准算法和肺算法ASIR、MBIRc、MBIRRP20和MBIRNR40重建,重建层厚0.625mm。在标准算法ASIR、MBIRc和MBIRNR40重建图像上测量胸廓入口层面、气管隆突下层面和肝门层面背部肌肉、皮下脂肪相同部位ROI的CT值与噪声(SD),并计算SNR,采用单因素方差分析比较各重建算法SD和SNR。于肺窗ASIR、MBIRc、MBIRRP20和纵隔窗标准算法ASIR、MBIRc、MBIRNR40进行噪声和细节结构清晰度5分法主观评分,并采用Wilcoxon符号等级检验进行统计学分析。结果初检有效剂量为(3.01±1.89)mSv,复查有效剂量为(0.88±0.83)mSv,下降约70.76%。MBIRNR40图像噪声明显低于常规剂量ASIR、低剂量ASIR和MBIRc(P均0.05)。MBIRNR40图像SNR绝对值明显大于常规剂量ASIR、低剂量ASIR和MBIRc(P均0.05)。MBIRNR40的主观图像噪声评分低于常规剂量ASIR和MBIRc(P均0.05);MBIRn可更清晰地显示肺、纵隔及上腹部细节结构,评分高于MBIRc和ASIR(P0.05)。结论在胸部CT平扫时,与ASIR、MBIRc相比,MBIRn肺特异性设置中MBIRNR40可显著降低图像噪声并提高SNR,可减少辐射剂量约70%,在低剂量条件下,MBIRRP20可更好地显示肺内、MBIRNR40可更好地显示纵隔、上腹部细节结构。
[Abstract]:Objective to compare the effects of self-adaptive statistical iterative reconstruction (ASIRR), conventional model-based iterative reconstruction (MBIRC), and new generation model-based iterative reconstruction (MBIRIRN) on the quality of sub#en0# chest CT images. Methods the lung specificity settings of MBIRRP20 and MBIRNR40) were compared. A total of 30 patients underwent two plain chest CT scans. The initial examination was performed with conventional dose (noise index 14) ASIR reconstruction. A low radiation dose scheme (noise index 28) was used. Standard and lung algorithms ASIRMBIRcI / MBIRRP20 and MBIRNR40 were used, respectively. The thickness of the reconstructed layer was 0.625mm. back muscles were measured on the standard algorithm ASIRN MBIRc and MBIRNR40 reconstruction images, including the thoracic entrance, the subtracheal protuberance and the hepatic hilar. The CT value and noise of ROI in the same area of subcutaneous fat were calculated, and the SD and SNR reconstruction algorithms were compared by single factor ANOVA. The noise and detail structure articulation were evaluated by 5 grades of noise and detail structure definition in lung window and mediastinal window standard algorithm ASIRM MBIRC / MBIRRP20 and the standard algorithm ASIR / MBIRcM / MBIRNR40 respectively, and compared with each other by single factor analysis of variance (ANOVA). The results showed that the initial effective dose was 3.01 卤1.89 mSvand the re-examination effective dose was 0.88 卤0.83 mSv.The image noise of MBIRNR40 decreased about 70.76. MBIRNR40 was obviously lower than that of the conventional dose. The absolute value of SNR in low dose ASIR and MBIRc(P were 0.05g 路MBIRNR40. The subjective image noise scores of low dose ASIR and MBIRc(P were lower than those of conventional dose ASIR and MBIRc(P both 0.05 and 0. 05 mb IRN could show the lung more clearly at the conventional dose of ASIR and MBIRc(P, and the subjective image noise score of MBIRNR40 was lower than that of normal dose ASIR and MBIRc(P. The detailed structure of mediastinum and epigastrium was higher than that of MBIRc and ASIRN (P 0.05). Compared with ASIRIRN lung specificity setting, MBIRNR40 can significantly reduce image noise and increase SNR, and reduce radiation dose by about 70%. MBIRRP20 can better display the mediastinal and epigastric details in the lung.
【作者单位】：陕西中医药大学附属医院医学影像科;
【分类号】：R816.4

【相似文献】