首页|不同重建算法对宝石能谱成像定量参数影响的体模研究

不同重建算法对宝石能谱成像定量参数影响的体模研究

Imaging quality of gemstone spectral imaging with different reconstruction algorithm using a phantom

扫码查看
目的:比较不同重建算法对宝石能谱成像(GSI)参数的影响,探讨深度学习图像重建(DLIR)在GSI碘基物质浓度(IC)和有效原子序数(Zeff)定量应用中的可行性和价值.方法:使用GSI模式扫描静止状态容纳9支试管的QSP体模,目标试管为1支纯水和5支碘溶液(碘浓度3.75、7.5、15 mg/mL各1支,30 mg/mL碘溶液2支).原始扫描数据分别使用标准算法0~100%基于多模型的迭代重建算法(ASIR-V)(间隔10%)和低、中、高权重3种深度学习图像重建(DLIR-L、DLIR-M和DLIR-H)层厚0.625 mm的IC和Zeff图像,并进行对比分析.选取9个固定层面在试管中心放置ROI,测量并比较不同重建算法图像的IC、Zeff均数及其标准差(SD).比较位于体模中心位置和边缘的2支碘浓度为30 mg/mL试管不同算法条件下的IC及Zeff.结果:不同算法纯水及各溶液中IC均能较准确测量碘含量,且差异均无统计学意义(均P>0.05).IC的SD方面,在纯水及不同浓度碘溶液中,DLIR-L和ASIR-V 40%~70%重建差异均无统计学意义(均P>0.05),DLIR-M和ASIR-V 70%~100%重建差异均无统计学意义(均P>0.05);余两两比较,差异均有统计学意义(均P<0.05).30 mg/mL碘溶液在体模中心位置IC略低于边缘位置,减小约0.63%.不同重建算法的Zeff均数差异均无统计学意义(均P>0.05).纯水和3.75 mg/mL碘溶液Zeff的SD方面,DLIR-L和ASIR-V 40%~70%重建差异均无统计学意义(均P>0.05);DLIR-M和ASIR-V 50%~90%、DLIR-L、DLIR-H重建差异均无统计学意义(均P>0.05);余各算法重建Zeff图上两两比较,差异均有统计学意义(均P<0.05).7.5、15、30 mg/mL碘溶液Zeff的SD方面,各种重建算法差异均无统计学意义(均P>0.05).30 mg/mL碘溶液在体模中心位置的Zeff 略低于边缘位置,减小约0.13%.结论:相较于ASIR-V,DLIR重建算法不改变IC和Zeff,但可在一定程度上减少噪声,其中DLIR-H噪声最低;扫描范围内ROI的位置改变对ASIR-V、DLIR重建IC和Zeff影响甚微.
Objective:To compare the effect of different reconstruction algorithms of gemstone spectral imaging(GSI)parameters,and to explore the feasibility and value of deep learning image reconstruction(DLIR)in quantifying iodine concentration(IC)and effective atomic number(Zeff).Methods:GSI mode was used to scan the QSP phantom which containing 9 test tubes,including 1 tube of pure water and 5 tubes of iodine solution with the iodine concentrations of 3.75,7.5,15,30 and 30 mg/mL.The original scanning data were reconstructed using the standard algorithm ASIR-V(ranged from 0%to 100%with the interval of 10%)and the high/middle/low-strength DLIR(DLIR-L/M/H)with a layer thickness of 0.625 mm for IC and Zeff images.ROIs were placed at the center of the test tubes at 9 fixed layers.The values of IC,Zeff and their average value and standard deviation(SD)of images from different reconstruction algorithms were measured and compared.The values of IC and Zeff under different algorithmic conditions of two tubes of iodine solution with iodine concentration of 30 mg/mL located at the center and edge of the phantom were compared.Results:Different algorithms images were able to be used for the measurements of iodine concentration in pure water and various solutions using IC,and there were no statistically significant differences(all P>0.05).In terms of the SD values of IC,there were no statistically significant differences between DLIR-L and ASIR-V(ranged from 40%to 70%)or between DLIR-M and ASIR-V(ranged from 70%to 100%)of pure water or different iodine solution(all P>0.05),while the remaining SD values showed statistical differences(all P<0.05).The center position IC of the phantom was slightly lower than the edge position,reducing by about 0.63%.Zeff values had no statistically significant differences of different reconstruction algorithms(all P>0.05).In terms of the SD values of Zeff in pure water and iodine solution with iodine concentration of 3.75 mg/mL,there were no statistically significant differences between DLIR-L and ASIR-V(ranged from 40%to 70%)or between DLIR-M and ASIR-V(ranged from 50%to 90%),DLIR-L,DLIR-H(all P>0.05),while the remaining SD values showed statistical differences(all P<0.05).There were no statistically significant differences in the SD values of Zeff of iodine solution with iodine concentration of 7.5,15 or 30 mg/mL in different reconstruction algorithms(all P>0.05).The center position Zeff of the phantom was slightly lower than the edge position,reducing by about 0.13%.Conclusions:Compared to ASIR-V,DLIR algorithm does not change IC and Zeff,but can reduce the noise to some extent,and DLIR-H has the lowest degree of noise.The position change has little effect on IC and Zeff in ASIR-V and DLIR.

Tomography,X-ray computedGemstone spectral imagingIodine concentrationEffective atomic number

贾永军、于楠、张喜荣、于勇、贺太平、段海峰、杨创勃

展开 >

陕西中医药大学附属医院医学影像科,陕西 咸阳 712000

体层摄影术,X线计算机 能谱成像 碘浓度 有效原子序数

陕西省教育厅青年创新团队科研计划项目陕西中医药大学创新团队基金项目陕西中医药大学校级科研课题国家基金面上培育项目

23JP0352019-YS042020GP10

2024

中国中西医结合影像学杂志
中国中西医结合学会,山东中医药大学附属医院

中国中西医结合影像学杂志

CSTPCD
影响因子:0.857
ISSN:1672-0512
年,卷(期):2024.22(4)
  • 11