基于模型的完全迭代算法(IMR)对肾脏小结石诊断价值的实验研究  

The Evaluation of a Knowledge-Based Iterative Model Reconstruction Algorithm for Small Kidney Calculi by Experimental Study

作　　者：田娟 沙琳 赖声远 刘书峰 

机构地区：[1]大连医科大学附属第二医院

出　　处：《临床放射学杂志》2017年第9期1359-1363,共5页Journal of Clinical Radiology

摘　　要：目的通过实验对比研究,探讨基于模型的完全迭代算法(IMR)技术对肾脏小结石的诊断价值。方法制作肾结石体模:单个重量(300±30)g的猪肾4个,将每个肾脏自肾脏一极沿冠状面切开至肾盂水平,将大小分别为2 mm、1 mm、0.5 mm的尿酸结石各24枚(每个肾脏6枚)分别植入肾皮质、肾髓质及肾盂,将植入结石的肾脏放置于注满生理盐水圆柱形有机容器,对体模进行CT扫描,用滤波反投影法(FBP)、高级迭代重组算法(iDose^4)、IMR技术分别重组出层厚为0.8 mm、层间距为0.3 mm的图像,对不同尺寸结石诊断的敏感性、特异性、准确性、数量、诊断结石的信心及图像质量进行评估。结果主观评价指标:(1)IMR图像的质量评分相对于FBP及iDose^4图像,有统计学意义。(2)IMR图像的噪声低于FBP及iDose^4图像,且iDose^4图像的噪声低于FBP。(3)对于2 mm、1 mm的结石三种重组方法对结石的检出率(均为100%)及正确诊断的信心(均为3分)没有统计学意义。对于0.5 mm的结石IMR技术对结石的检出率及正确诊断的信心与FBP和iDose^4比较,P<0.05,有统计学意义。客观评价指标:FBP、iDose^4、IMR技术所得图像的噪声和信噪比分别为[(5.8±0.5)HU,9.1±0.6]、[(4.1±0.3)HU,13.5±0.5]、[(1.9±0.3)HU,28.2±0.5],三组图像的噪声和信噪比两两比较,差异均有统计学意义。结论相对于FBP与iDose^4图像,IMR技术能在提高图像质量的同时提高微小结石诊断的准确性。Objective To evaluate the efficacy of a knowledge-based iterative model reconstruction（ IMR） algorithm for small kidney calculi by experimental study. Methods All uric acid stones were collected from urology department.Those renal calculi were implanted into four pig kidneys. Single kidney weight 300 g ± 30 g,each kidney calyx from a cut along the coronal plane to the level of the renal pelvis. 2mm,1mm,0. 5mm diameter of uric acid stones was implanted into the renal cortex,medulla and renal pelvis respectively. The kidneys were scanned in a cylinder-shaped glass container,which is filled with 0. 9% Na Cl solution,while the scanning tube voltage is 100 k V and the tube current is 100 m A. Using filtered back-projection（ FBP） 、iDose4、IMR reconstructed thickness of 0. 8mm,0. 3mm layer spacing image respectively.Two experienced doctors then evaluated the images,including the diagnostic sensitivity,specificity,accuracy,the number of calculi,calculi diagnostic confidence and image quality. Results Subjective Evaluation： 1. IMR image quality ratings relative to the FBP and iDose images,showed that there was statistically significant difference. 2. IMR image noise is lower than the FBP and IMR,and IMR image noise is lower than FBP too. 3. For 2mm,1mm stones three kinds of reconstruction methods of calculus detection rate（ 100%） and correct diagnosis of confidence（ 3 points in both） were not statistically significant.For 0. 5mm stones of IMR technology of correct diagnosis confidence and calculi detection compared with FBP and iDose4,P〈 0. 05,the difference was statistically significant. Objective evaluation： The noise and signal to noise ratio of FBP image、iDose4 image,MR image were respectively [（ 5. 8 ± 0. 5） HU,9. 1 ± 0. 6]、[（ 4. 1 ± 0. 3） HU,13. 5 ± 0. 5]、[（ 1. 9 ±0. 3） HU,28. 2 ± 0. 5 ],three sets of image noise and SNR pairwise comparison,the differences were statistically significant. Conclusion Compared with FBP and iDose4 image,IMR technology can improve the acc

关 键 词：滤波反投影法 高级迭代重组算法 基于模型的完全迭代算法 结石 

分 类 号：R445.2[医药卫生—影像医学与核医学] R692.4[医药卫生—诊断学]