Multivendor Comparison of Quantification Accuracy of Iodine Concentration and Attenuation Measurements by Dual-Energy CT: A Phantom Study
Yong Chen,Jingyu Zhong,Lan Wang,Xiaomeng Shi,Rui Chang,Jing Fan,Jiang,Yihan Xia,Fuhua Yan,Weiwu Yao,Huan Zhang
DOI: https://doi.org/10.2214/ajr.22.27753
2022-01-01
Abstract:BACKGROUND. Studies comparing accuracy of quantification by dual-energy CT (DECT) scanners have been limited by small numbers of scanners evaluated and narrow ranges of scanning conditions. OBJECTIVE. The purpose of this study was to compare DECT scanners of varying vendors, technologies, and generations in terms of the accuracy of iodine concentration and attenuation measurements. METHODS. A DECT quality-control phantom was designed to contain seven inserts of varying iodine concentrations as well as soft-tissue and fat inserts. The phantom underwent DECT using 12 different scanner configurations based on seven different DECT scanners from three vendors, with additional variation in tube voltage settings. Technologies included rapid-switching, dual-source, and dual-layer detector DECT. Scans also used three radiation dose levels (10, 20, and 30 mGy) and multiple reconstruction algorithms (filtered back projection, medium and high iterative reconstruction, and deep learning image reconstruction [DLIR]). The mean absolute percentage error (MAPE, representing the absolute ratio of measured error to nominal values on average; lower values indicate better accuracy) was calculated for iodine concentration on iodine maps (MAPEiodine) and attenuation on virtual monochromatic images (VMIs) using 40, 70, 100, and 140 keV (MAPEHU). Linear mixed models were used to explore factors affecting quantification accuracy. RESULTS. MAPEiodine and MAPEHU ranged 4.62-28.55% and 10.21-26.33%, respectively, across scanner configurations. Accuracies of iodine concentration and attenuation measurements were higher for third-generation rapid-switching and dual-source scanners in comparison with respective earlier-generation scanners and the single evaluated dual-layer detector scanner. Among all configurations, the third-generation rapid-switching scanner using DLIR had the highest quantification accuracy for iodine concentration (MAPEiodine, 4.62% ± 3.87%) and attenuation (MAPEHU, 10.21% ± 11.43%). Overall, MAPEiodine was significantly affected by scanner configuration (F = 450.0, p < .001) and iodine concentration (F = 211.0, p < .001). Overall, MAPEHU was significantly affected by scanner configuration (F = 233.5, p < .001), radiation dose (F = 14.9, p < .001), VMI energy level (F = 1959.4, p < .001), and material density (F = 411.5, p < .001); radiation dose was significantly associated with MAPEHU for five of 12 individual configurations. CONCLUSION. Quantification accuracy varied among DECT configurations of varying vendors, platforms, and generations and was affected by acquisition and reconstruction parameters. DLIR may improve quantification accuracy. CLINICAL IMPACT. The interscanner differences in DECT-based measurements should be recognized when quantitative evaluation is performed by DECT in clinical practice.