迭代算法对CT低剂量扫描潜能的体模研究  

The phantom study of the potential of radiation dose reduction using adaptive iterative dose reduction algorithm

作　　者：缪妙[1] 曹国全[2] 邰云鹏[2] 曾碧新[3] 

机构地区：[1]温州医科大学附属第一医院医学工程处,浙江温州325015 [2]温州医科大学附属第一医院放射科,浙江温州325015 [3]温州医科大学信息与工程学院,浙江温州325035

出　　处：《温州医科大学学报》2015年第11期845-848,853,共5页Journal of Wenzhou Medical University

摘　　要：目的:研究在保证图像质量的前提下,自适应迭代降剂量算法(AIDR 3D)降低CT扫描剂量的潜能。方法:采用320层CT容积扫描模式,用不同管电流(50、80、150、250、350、450、550 m A)扫描体模Catphan500的低对比度模块CTP515和高分辨率模块CTP528,并用滤过反投影重建算法(FBP)与AIDR 3D的三个迭代等级(mild/standard/strong)分别重建图像,对比分析图像的噪声、对比噪声比、信噪比以及空间分辨率等图像质量参数。结果:采用AIDR 3D重建可以显著降低图像噪声,提高对比噪声比和信噪比,同时不改变空间分辨率。最高降低辐射比例达78.41%。结论:AIDR 3D迭代算法具有良好的降低CT扫描剂量的能力。不同的迭代等级降低剂量的潜能不同。Objective： To study the ability of the reconstruction algorithm of adaptive iterative dose reduction 3D to reduce the radiation dose of CT scanning, under the premise in image quality undamaged. Methods： A 320-detector CT with volume mode was used to scan the low contrast module CTP515 and the high resolution module CTP528 of the phantom Catphan 500, while seven different tube currents 50, 80, 150, 250, 350, 450, and 550 mA were adopted respectively. CT images were reconstructed using the reconstruction algorithm of Filtered Back Projection and AIDR 3D in three degrees： mild, standard and strong. The noise, contrast-noise ratio, signal- noise ratio and spatial resolution of the images were used to evaluate the quality of the output image yielded by different tube-ctLrrent/reconstruction-algorithm choices. Results： Using the algorithm ofAIDR 3D to reconstruct CT image can significantly reduce image noise and increase contrast-noise ratio and signal-noise ratio with the spatial resolution undamaged. In the best case, the proportion of the radiation could be reduced by 78.41%. Conclusion： The reconstruction algorithm ofAIDR 3D has remarkable ability to reduce the radiation dose of CT scanning. Different iteration degrees resultes in different level of reduction.

关 键 词：计算机断层 迭代重建 体层摄影术 X线计算机 辐射剂量 图像质量 

分 类 号：R814.42[医药卫生—影像医学与核医学]