Iodine concentration,HU accuracy,and metal artifacts evaluation on second-generation dual-layer spectral detector CT images with metal implants: a phantom study

Abstract

Background

Metal implants may affect the image quality, iodine concentration (IC), and CT Hounsfield unit (HU) quantification accuracy.

Purpose

To investigate the quantitative accuracy of IC and HU from dual-layer spectral detector (DLCT) in the presence of metal artifacts.

Material and Methods

An experimental cylindrical phantom containing eight iodine inserts and two metal inserts was designed. The phantom underwent scanning at three radiation dose levels and two tube voltage settings. A set of conventional images (CIs), virtual monoenergetic images (VMIs), and iodine concentration maps (ICMs) were generated and measured for all the eight iodine inserts. Quantitative indicators of mean absolute percentage error (MAPE), artifact index (AI), contrast-to-noise ratio (CNR), signal-to-noise ratio (SNR), and standard deviation (SD) on CIs and VMIs were calculated for IC and HU. Subjective score evaluation was also conducted.

Results

The MAPE_iodine values of all regions of interest across different scanning configurations were all <5%. Almost all APE_iodine values were <5%, indicating that metal artifacts had little impact on IC measurements. When the tube voltage was fixed, the SD value of attenuation decreased with the increase of the tube current; this is also true when the tube current was fixed. The middle energy reconstructions seemed to give a good balance between reducing artifacts and improving contrast.

Conclusion

VMIs from DLCT can reduce metal artifacts, the accuracy of IC quantification is not sensitive to imaging parameters. In summary, metal implants exhibit minimal impact on image quality and IC quantification accuracy in reconstructed images from DLCT.

Keywords

Dual-layer spectral detector iodine concentration phantom study virtual monoenergetic images

Introduction

Currently, three main types of dual-energy spectral computed tomography (DECT) scanners are widely used: dual-source CT scanners; fast-kV switching CT scanners; and two-layer spectral detector CT scanners (1,–6). With a single gantry, two X-ray tubes and detector array systems are used in dual-source CT scanners to simultaneously acquire low- and high-energy data with an offset of approximately 90° (1,2). Fast-kV switching CT scanners use a single X-ray tube and detector array, switching between low and high voltage in consecutive projection sets (3,4). As the first commercially available spectral CT system based on detectors, dual-layer spectral detector CT (DLCT) consists of two layers of scintillators. The upper layer is a Yttrium-based scintillator – its role is to absorb low-energy photons from the X-ray beam – and the lower layer is gadolinium oxide sulfide (GOS), which plays the role of absorbing high-energy photons from the top layer (5,6). The purpose of selecting appropriate scintillator thickness is to obtain the best energy separation and signal-to-noise ratio (SNR).

Phantom studies with inserts for material composition that are clearly described enable for accurate quantitative system evaluation across a variety of clinical objectives (3714). Several studies have been performed to evaluate the measurement accuracy of spectral CT reconstructions. For the fast KV-switching spectral CT system, iodine maps and virtual monoenergetic images (VMIs) were tested for correctness by Zhang et al. (3), and effective atomic numbers and VMIs were evaluated by Goodsitt et al. (7). Iodine quantification accuracy was tested between dual-source and two-layer detector spectral CT scanners by Pelgrim et al. (8).

In clinical practice, the image quality is easily affected by the artifacts produced by metal implants (15,16). These artifacts combine to create significant interferences that degrade the image quality of the metal implant and the tissues around it (teeth, bone, and soft tissue). In conventional single-energy CT imaging, different methods to reduce metal artifacts (MAs) have been proposed: high tube voltage (kVp); high tube current (mAs); modified collimation; and suitable filters for reconstruction algorithms (17). There are a number of postprocessing techniques for MA reduction (also known as MAR algorithms) that have been demonstrated to reduce metal artifacts (18,19). In addition, VMIs from DECT can be used for MA reduction (20,–22). As DECT evaluates the attenuation of photons with different energies individually, it enables the imitation reconstruction of VMIs at various energies. Higher keV VMIs are less prone to beam hardening and have demonstrated a reduction in MA from a range of metal implants (23,–27).

To the best of our knowledge, there are no studies on whether metal artifacts have a significant negative impact on the accuracy of iodine concentration (IC) of commercial dual-energy DLCT system. Therefore, the aim of the present study was to evaluate the accuracy of both IC and CT Hounsfield unit (HU) measurements within a phantom on spectral reconstructed images from DLCT with metal implants.

Material and Methods

Dual-lay detector CT test phantom

A quality-control phantom designed by our research team was used to analyze the IC and CT HU accuracy. The phantom consists of a solid cylindrical shape made of resin. The cylinder has a diameter of 20 cm, a thickness of 15 cm, and is filled with distilled water. The 10 holes on the phantom enable the placement of different material inserts for quantitative measurement. The inserts measure 2.5 cm in diameter and 10 cm in depth. A total of eight iodine inserts with ICs of 2.0, 2.5, 5.0, 7.5, 10.0, 12.5, 15.0, and 20.0 mg/mL and two additional metal material inserts were inserted into the remaining two holes of the phantom to evaluate iodine quantification accuracy and the effectiveness of VMIs on metal artifact correction. Further details of phantom configuration are shown in Fig. 1.

Fig. 1.

Phantom configuration. The quality-control phantom in the CT image had a diameter of 20 cm and eight iodine inserts with concentrations in the range of 2.0–20.0 mg/mL (increasing in clockwise orientation). Using each of the eight iodine inserts, attenuation measurements were calculated (dotted rings). Two additional inserts numbered 1 and 2 (solid rings) correspond to metal materials used for metal artifact evaluation.

Scanner configurations and DECT acquisitions

We evaluated one second-generation DLCT scanner (Spectral CT; Philips Healthcare, Guangzhou, PR China) installed in the clinical department of Nanfang Hospital in China. To reach CTDIvol values as close as feasible to 10, 20, and 30 mGy, this scanner acquired images while operating at two different tube voltage settings (120 and 140 kVp) and three different radiation doses. We can achieve this by adjusting the tube current, pitch, and rotation time. Therefore, we used six configurations for the combination of tube voltage setting and radiation dose level. A procedure created to mimic a clinical abdominopelvic CT routine was used to conduct scans (e.g. radiation doses similar to those of abdominopelvic CT examinations and reconstruction using soft-tissue kernels). Other clinical experimental parameters were as follows: field of view = 50 × 50 cm; rotation time = 2.44 s; pitch = 0.99; matrix = 512 × 512; reconstruction thickness = 5 mm; increment = 5 mm; adopting automatic exposure control. For each combination of tube voltage setting and radiation dose, the CT acquisition was carried out three times while holding all other variables constant. A set of spectral-based image (SBI) data was then generated from the console for each scanning configuration.

DECT image analysis

For each scanning configuration, conventional images (CIs), VMIs in the range of 40–200 keV, and iodine concentration maps (ICMs) were reconstructed from the SBI data and analyzed using a dedicated workstation (IntelliSpace Portal, version 10.1; Philips Healthcare). CIs were reconstructed using iterative reconstruction algorithm (iDose4, level 4; Philips Healthcare). VMIs in the range of 40–200 keV with intervals of 10 keV and ICMs were reconstructed by spectral algorithm (spectral reconstruction, level 4; Philips Healthcare). Quantitative measures were made by an abdominal radiologist with seven years of imaging postprocessing experience. For the eight iodine inserts, the radiologist placed eight single circular regions of interests (ROIs) with an area of approximately 90 mm² on the slice with visually maximum artifacts on CIs, respectively, and then pasted the ROIs on VMIs and ICMs with the same location since all the images were reconstructed from the same SBI data for each scanning configuration. The mean ± standard deviation (SD) of each ROI was recorded.

Accuracy of IC measurements

As the measurements on all the images scanned without metal inserts were regarded as the reference, the accuracy of IC measurements with two metal inserts was assessed by comparing the values with those values without metal inserts for the eight ROIs. The following formula was applied to each scanning configuration to get the mean absolute percentage error (MAPE) of the IC measurements (MAPE_iodine):

M A P E_{i o d i n e} = \frac{1}{n} \sum_{i = 1}^{n} \frac{| {\hat{I C}}_{i} - I C_{i} |}{I C_{i}} \times 100 %

(1)

where

I C_{i}

and

{\hat{I C}}_{i}

represent the measured IC without metal on CT image and the measured IC with metal in CT image, respectively, and n represents the number of measurements. A lower MAPE_iodine value indicates more accurate quantification of IC.

Accuracy of attenuation measurements

Based on the measurements from the eight iodine inserts, the accuracy of attenuation measures was evaluated and determined independently for each VMI energy level. In a manner similar to that used to assess the accuracy of IC, for each arrangement, the MAPE of attenuation (MAPE_HU) was determined using the following formula:

M A P E_{H U} = \frac{1}{n} \sum_{i = 1}^{n} \frac{| {\hat{H U}}_{i} - H U_{i} |}{H U_{i}} \times 100 %

(2)

where

H U_{i}

and

{\hat{H U}}_{i}

represent the measured iodine attenuation without metal in CT image and the measured iodine attenuation with metal in CT image, respectively, and n represents the number of measurements. More accurate attenuation quantification was indicated by a lower MAPE_HU.

Evaluation of metal artifacts reduction

The evaluation of metal artifacts reduction was based on measurements from eight iodine inserts and determined separately for each energy level of VMIs. The following formulas were used to determine the SNR and contrast-to-noise ratio (CNR) for each configuration:

A I = \frac{1}{n} \sum_{i = 1}^{n} (| {\hat{S D}}_{i}^{2} - S D_{i}^{2} |)

(3)

S N R = \frac{1}{n} \sum_{i = 1}^{n} \frac{{\hat{H U}}_{i}}{{\hat{S D}}_{i}}

(4)

C N R = \frac{1}{n} \sum_{i = 1}^{n} \frac{| {\hat{H U}}_{i} - H U_{i} |}{\sqrt{S D_{i}^{2} + {\hat{S D}}_{i}^{2}}}

(5)

where

H U_{i}

and

S D_{i}

represent the measured ICs and standard deviation without metal in CT image, the

{\hat{H U}}_{i}

and

{\hat{S D}}_{i}

represent the measured ICs and standard deviation with metal in CT image, and n is the quantity of measurements, respectively. Lower AI and higher SNR and CNR indicated fewer metal artifacts effects and more accurate quantification.

In addition, the subjective score was also used to evaluate metal artifacts. Specifically, the scores were scaled by 1, 2, 3, and 4 as follows: 1 = severe artifacts; 2 = middle artifacts; 3 = mild artifacts; and 4 = no artifacts. Two radiologists independently conducted the subjective assessment.

Statistical analysis

For each scanning configuration, the mean ± SD of MAPE_iodine and MAPE_HU values were calculated across inserts, radiation doses, and voltage settings, as well as MAPE_HU's energy level, to provide a total figure for comparison. IC (with/without metal on images), radiation dosage (i.e. CTDI_vol), and tube voltage settings were all fixed effects in the model used to evaluate IC accuracy. The VMIs were included as a fixed effect in the model that was used to evaluate attenuation accuracy. The dependent variable in each model was either MAPE_iodine or MAPE_HU. The numerical variables were compared using the Kruskal–Wallis and Mann–Whitney tests. The inter-reader agreement of the subjective rating of the image was evaluated using the intraclass correlation coefficient (ICC). The cutoff for statistical significance was P < 0.05. SPSS software version 26 (IBM Corp., Armonk, NY, USA) and R software version 4.0.3 (R Foundation, Vienna, Austria) were used to conduct the statistical analysis.

Results

Accuracy of IC measurements

In general, the MAPE_iodine values across different scanning configurations for both with/without metal inserts were all below 5% (Table 1), which was acceptable in clinic, and the P values using the Kruskal–Wallis test were 0.78 and 0.20, respectively. For the case of MAPE_iodine (with metal), there was no statistical difference using the Kruskal–Wallis test when the voltage was fixed and the current varied (120 kV: P = 0.263; 140 kV: P = 0.712); There was a statistical difference using the Mann–Whitney test when 30 mGy was fixed and the voltage varied (10 mGy: P = 0.62; 20 mGy: P = 0.375; 30 mGy: P = 0.023).

Table 1.

MAPE_iodine values (in %) for each scanning configuration.

Configurations	MAPE_iodine (no metal)	MAPE_iodine (with metal)
120 kV/10 mGy	2.96 ± 2.26	2.10 ± 1.28
120 kV/20 mGy	3.17 ± 3.14	1.96 ± 1.73
120 kV/30 mGy	3.29 ± 2.78	2.39 ± 3.57
140 kV/10 mGy	4.09 ± 3.86	2.77 ± 3.73
140 kV/20 mGy	3.12 ± 2.21	2.40 ± 3.52
140 kV/30 mGy	3.03 ± 2.51	1.43 ± 2.07
P value	0.78	0.20

Values are given as mean ± SD.

MAPE, mean absolute percentage error; SD, standard deviation.

Aside from MAPE_iodine, Table 2 showed results in terms of absolute percentage error (APE_iodine) for each ROI (iodine insert) in each scanning configuration. In ROI1, most APE_iodine values were above 5%, while nearly all APE_iodine values from ROI2 to ROI8 were below 5%, respectively. Table 2 also shows that, in general, the APE_iodine value of concentration measurements in each ROI with different configurations decreased as the nominal IC increased.

Table 2.

APE_iodine values (in %) for each ROI in each scanning configuration.

	ROI index	120 kV/10 mGy	120 kV/20 mGy	120 kV/30 mGy	140 kV/10 mGy	140 kV/20 mGy	140 kV/30 mGy
APE_iodine(with metal inserts)	ROI1	1.80 ± 0.81	3.20 ± 0.66	11.70 ± 0.00	11.70 ± 1.04	11.30 ± 0.24	6.70 ± 0.44
	ROI2	3.20 ± 0.32	5.50 ± 0.20	1.10 ± 0.20	3.80 ± 0.38	3.60 ± 0.19	0.40 ± 0.00
	ROI3	1.60 ± 0.24	2.90 ± 0.18	0.80 ± 0.00	4.30 ± 0.24	1.30 ± 0.09	1.90 ± 0.19
	ROI4	3.10 ± 0.06	0.30 ± 0.13	2.00 ± 0.00	0.40 ± 0.06	0.60 ± 0.06	0.20 ± 0.06
	ROI5	4.00 ± 0.41	1.80 ± 0.13	1.60 ± 0.13	1.20 ± 0.81	1.50 ± 0.09	1.20 ± 0.17
	ROI6	2.20 ± 0.07	1.00 ± 0.04	0.70 ± 0.04	0.40 ± 0.09	0.70 ± 0.03	0.50 ± 0.09
	ROI7	0.70 ± 0.08	0.10 ± 0.08	0.40 ± 0.09	0.20 ± 0.06	0.20 ± 0.05	0.20 ± 0.13
	ROI8	0.10 ± 0.06	0.80 ± 0.10	0.70 ± 0.00	0.20 ± 0.06	0.10 ± 0.00	0.40 ± 0.02

Values are given as mean ± SD.

APE, absolute percentage error; ROI, region of interest; SD, standard deviation.

Accuracy of attenuation measurements

Fig. 2 presented the MAPE_HU values in different energy levels of VMIs and CIs. For CIs, 20 mGy with different voltages had the smallest MAPE_HU. There was no statistical difference using the Kruskal–Wallis test for CIs with all scanning configurations (P = 0.298). The MAPE_HU values on VMIs at 40–80 keV were all lower than those on CIs. MAPE_HU below 5% could be seen on VMIs at 40–100 keV. The configuration of 140 kV/30 mGy obtained better results in comparison with other configurations. Among results in configuration of 140 kV/30 mGy, VMIs at 80 keV got the smallest MAPE_HU value. Finally, among all VMIs, MAPE_HU climbed up with the increasing energy levels and the MAPE_HU corresponding to the configuration (120 kV/20 mGy) changed dramatically.

Fig. 2.

MAPE_HU values in different VMIs and conventional images (CIs).

Table 3 presented the SD values for each ROI in each scanning configuration. Among eight ROIs, ROI5 showed the highest SD values for the six scanning configurations. In general, the SD values decreased as the radiation dose rose while the tube voltage was fixed, and the opposite was true when the radiation dose was fixed.

Table 3.

Values for each ROI (in HU) in each scanning configuration.

	120 kV/10 mGy	120 kV/20 mGy	120 kV/30 mGy	140 kV/10 mGy	140 kV/20 mGy	140 kV/30 mGy
ROI1	13.10 ± 0.16	8.93 ± 0.09	6.07 ± 0.29	10.07 ± 0.68	6.67 ± 0.05	6.77 ± 0.09
ROI2	11.50 ± 0.00	6.47 ± 0.05	5.20 ± 0.08	9.67 ± 0.05	6.03 ± 0.05	5.43 ± 0.09
ROI3	9.30 ± 0.08	6.17 ± 0.05	4.70 ± 0.08	9.40 ± 0.43	5.37 ± 0.05	5.13 ± 0.05
ROI4	9.10 ± 0.16	6.53 ± 0.05	5.17 ± 0.05	9.40 ± 0.68	6.63 ± 0.09	6.33 ± 0.05
ROI5	16.53 ± 0.31	12.20 ± 0.08	12.33 ± 0.05	12.23 ± 2.22	12.07 ± 0.09	12.10 ± 0.16
ROI6	9.50 ± 0.22	6.80 ± 0.08	6.50 ± 0.08	10.27 ± 0.05	7.87 ± 0.05	4.80 ± 0.00
ROI7	9.43 ± 0.12	8.10 ± 0.08	6.83 ± 0.17	8.23 ± 0.46	6.83 ± 0.09	5.13 ± 0.29
ROI8	13.10 ± 0.00	8.93 ± 0.19	6.07 ± 0.05	10.07 ± 0.05	6.67 ± 0.22	6.77 ± 0.12
Overall	11.08 ± 2.43	8.02 ± 2.43	6.721 ± 2.58	9.88 ± 2.37	7.36 ± 2.37	6.53 ± 2.06

Values are given as mean ± SD.

HU, Hounsfield unit; ROI, region of interest; SD, standard deviation.

Evaluation of metal artifacts reduction

Fig. 3 shows different evaluation indexes of metal artifacts reduction for varying scanning configurations. Fig. 3A shows the AI evaluation. VMIs at 40 keV had the highest AI values compared to other VMIs. In each group of VMIs, 120 kV/10 mGy had the highest value of AI. With the intensity of the energy level, the AI values tended to decrease and become stable after 120 keV. Fig. 3B shows the CNR evaluation. In the 120 kV/30 mGy configuration, when the energy level increased, the CNR decreased dramatically before 90 keV and then tended to be constant. In addition, after 90 keV, the CNR in the other five scanning configurations rose slightly. Fig. 3C showed the SNR evaluation. The 140 kV/30 mGy configuration had the highest SNR for each level of VMIs and had a tendency of decreasing SNR when the energy level increased. Fig. 3D showed the SD of HU evaluation, similar to Fig. 3A. When the energy level increased, the SD values for different scanning configurations dropped to 90 keV. After that, the SD remained constant.

Fig. 3.

Different evaluation indexes of metal artifacts reduction for varying scanning configurations: (A) AI; (B) CNR; (C) SNR; and (D) SD. AI, artifact index; CNR, contrast-to-noise ratio; SD, standard deviation; SNR, signal-to-noise ratio.

Fig. 4 shows the subjective evaluation in CI and VMI_{40 keV–200 keV}. We found that different scanning configurations had similar results to the subjective assessment, and the agreement between the two readers was excellent (ICC = 0.85). Fig. 4 presents the average results for all configurations. It can be seen that VMIs were rated superior compared with CI except for VMI_40 keV, and the best score could be achieved after VMI_120 keV.

Fig. 4.

Results of the subjective assessment.

Discussion

The present study confirmed that the accuracy of iodine quantification and attenuation measures remains robust across varying scanning protocols in the presence of metal artifacts using DLCT. Notably, IC measurements were not significantly affected by different scanning configurations for cases both with and without metals. The accuracy of attenuation measurement in VMIs at 40–80 keV were superior to those on CIs. Images at the medium energy reconstructions were regarded as having better diagnostic image quality because they appeared to strike a solid balance between contrast improvement and artifact removal.

At the time of writing, only a little amount of research has been carried out on how well DLCT performed when metal implants were present on the DLCT images. Some previous research on spectral CT showed that MAR algorithm and high-keV VMIs might significantly lessen the artifacts brought on by metal implants. Other spectral CT research presented that DLCT could be used to precisely measure even low iodine amounts (28). Yet the quantitative precision of iodine assays was not examined in these studies and attenuation measurements using a phantom with iodine inserts and metal inserts. Our investigation addressed this gap, focusing on the precision of iodine quantification and attenuation measurements in the presence of metal artifacts. The varying ICs used in our study provided a comprehensive evaluation of iodine quantification accuracy with metal implants, shedding light on the challenges and potential solutions.

Our results showed that tube voltage settings (120 and 140 kVp) had no effect on the precision of iodine quantification (MAPE_iodine values) and three radiation doses settings (10, 20, and 30 mGy) and all the MAPE_iodine values were less than 5%, which are acceptable in clinic. In addition, almost all APE_iodine values from ROI2 to ROI8 were below 5%, indicating that metal artifacts had little impact on IC measurements. Several factors may help to explain this disparity. Since ROI1 included the lowest IC among the eight ROIs, the measured IC was easily affected by varying factors, such as noise, artificial errors, design errors, and so on. High-performance liquid chromatography is needed to make exact measurements of the true iodine content to resolve this problem. Second, the metal implants in the phantom might have a negative impact of IC measurement due to the metal artifacts. For patients with metal implants, the measurement of APE_iodine can be used as a reference to establish a real change in iodine. We think that the core data from our experiment supports the use of iodine quantification in metal artifact characterization.

On different VMIs, the MAPE_HU below 5% could be seen on VMIs at 40–100 keV images. Therefore, it is recommended that radiologists prioritize low-energy images (40–100 keV) when focusing on attenuation accuracy and opt for high-energy images when considering the reduction of metal artifacts. We think that in the future, CT protocol settings may take our findings into account as potential supporting data for patients who have metal implants.

The SD values within the ROIs were measured to assess the inhomogeneity of the CT value. ROI5 showed the largest SD values among the eight ROIs. This issue might be explained by the following reason: the disturbance generated by metal implants may be influenced by gravity, and ROI5 was located at the bottom of phantom, which resulted in the largest SD value in ROI5. Thus, the tissue located beneath the metal implants maybe have the most serious artifacts. But it will need more clinical practice to validate. Furthermore, as the tube voltage and current increased, we discovered that the SD values decreased.

We compared metal artifacts on CIs with those on VMIs using different parameters including AI, CNR, SNR, SD, and subjective score (22,27). Metal artifacts were found on VMIs with a considerable objective artifact reduction. This was consistent with earlier research that claimed a drop in contrast and increase in subjective score with increasing energy level of VMIs (22). We found the optimal overall diagnostic image quality on the middle energy reconstructions (80–110 keV). This appeared to strike a compromise between contrast enhancement and artifact removal. Because beam hardening artifacts are caused by modifications in the beam spectrum as it is attenuated and dual-energy CT can characterize the transmitted beam energy, VMIs may be able to reduce these artifacts. Receiving a predetermined reconstruction of the VMIs that is tailored for metal artifacts would be ideal for the radiologists.

The present study has some limitations. To begin with, we did not assess the iodine measurements and attenuation measurements in patients, which did not represent real-world clinical conditions. Second, we did not test the scanner with various phantom sizes. Although not the main topic of this investigation, the impact of phantom size may be covered in further work. Third, the impact of patient motion was beyond the study's scope, but further research should be done on this intriguing subject.

In conclusion, this study emphasizes the imaging configuration and IC had no discernible impact on the quantification accuracy of the iodine measurements, and metal implants do not affect the accuracy of iodine and HU quantification in DLCT. When using DECT in clinical practice for quantitative evaluation, it is important to be aware of these inter-scanner variations in measurements. Further research interest will be focused on the use of specialized metal artifact removal methods on virtual monoenergetic reconstructions and study its effect on IC quantification accuracy.

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Xinming Zhang

References

Johnson

Krauss

Sedlmair

, et al. Material differentiation by dual energy CT: initial experience. Eur Radiol 2007;17:1510–1517.

Schmidt

Flohr

. Principles and applications of dual source CT. Phys Med 2020;79:36–46.

Zhang

Liu

. Objective characterization of GE discovery CT750 HD scanner: gemstone spectral imaging mode. Med Phys 2011;38:1178–1188.

Rajiah

Parakh

Kay

, et al. Update on multienergy CT: physics, principles, and applications. RadioGraphics 2020;40:1284–1308.

Romman

Uman

Yagil

, et al.Detector technology in simultaneous spectral imaging. Philips White Paper, available online 2015, 29.

Rassouli

Etesami

Dhanantwari

, et al. Detector-based spectral CT with a novel dual-layer technology: principles and applications. Insights Imaging 2017;8:589–598.

Goodsitt

Christodoulou

Larson

. Accuracies of the synthesized monochromatic CT numbers and effective atomic numbers obtained with a rapid kVp switching dual energy CT scanner. Med Phys 2011;38:2222–2232.

Pelgrim

van Hamersvelt

Willemink

, et al. Accuracy of iodine quantification using dual energy CT in latest generation dual source and dual layer CT. Eur Radiol 2017;27:3904–3912.

Jacobsen

Schellingerhout

Wood

, et al. Intermanufacturer comparison of dual-energy CT iodine quantification and monochromatic attenuation: a phantom study. Radiology 2018;287:224–234.

10.

Dabli

Frandon

Hamard

, et al. Optimization of image quality and accuracy of low iodine concentration quantification as function of kVp pairs for abdominal imaging using dual-source CT: a phantom study. Phys Med 2021;88:285–292.

11.

Dabli

Frandon

Belaouni

, et al. Optimization of image quality and accuracy of low iodine concentration quantification as function of dose level and reconstruction algorithm for abdominal imaging using dual-source CT: a phantom study. Diagn Interv Imaging 2022;103:3–40.

12.

Kim

Goo

Kang

, et al. Comparison of iodine density measurement among dual-energy computed tomography scanners from 3 vendors. Invest Radiol 2018;53:321–327.

13.

Sellerer

Noël

Patino

, et al. Dual-energy CT: a phantom comparison of different platforms for abdominal imaging. Eur Radiol 2018;28:2745–2755.

14.

Euler

Solomon

Mazurowski

, et al. How accurate and precise are CT based measurements of iodine concentration? A comparison of the minimum detectable concentration difference among single source and dual source dual energy CT in a phantom study. Eur Radiol 2019;29:2069–2078.

15.

De Crop

Casselman

Van Hoof

, et al.

Analysis of metal artifact reduction tools for dental hardware in CT scans of the oral cavity: kVp iterative reconstruction, dual-energy CT, metal artifact reduction software: does it make a difference?

Neuroradiology 2015;57:841–849.

16.

Cha

Kim

H-J

Kim

, et al. Dual-energy CT with virtual monochromatic images and metal artifact reduction software for reducing metallic dental artifacts. Acta Radiol 2017;58:1312–1319.

17.

Lee

M-J

Kim

Lee

S-A

, et al. Overcoming artifacts from metallic orthopedic implants at high-field-strength MR imaging and multi-detector CT. Radiographics 2007;27:791–803.

18.

Andersson

Nowik

Persliden

, et al. Metal artefact reduction in CT imaging of hip prostheses-an evaluation of commercial techniques provided by four vendors. Br J Radiol 2015;88:20140473.

19.

Kidoh

Nakaura

Nakamura

, et al. Reduction of dental metallic artefacts in CT: value of a newly developed algorithm for metal artefact reduction (O-MAR). Clin. Radiol 2014;69:e11–e16.

20.

Flohr

McCollough

Bruder

, et al. First performance evaluation of a dual-source CT (DSCT) system. Eur Radiol 2006;16:256–268.

21.

Lewis

Reid

Toms

. Reducing the effects of metal artefact using high keV monoenergetic reconstruction of dual energy CT (DECT) in hip replacements. Skeletal Radiol 2013;42:275–282.

22.

Wellenberg

Boomsma

van Osch

, et al. Quantifying metal artefact reduction using virtual monochromatic dual-layer detector spectral CT imaging in unilateral and bilateral total hip prostheses. Eur J Radiol 2017;88:61–70.

23.

McCollough

Leng

, et al. Dual- and multi-energy CT: principles, technical approaches, and clinical applications. Radiology 2015;276:637–653.

24.

Johnson

TRC

. Dual-energy CT: general principles. AJR Am J Roentgenol 2012;199:S3–S8.

25.

Alvarez

Macovski

. Energy-selective reconstructions in X-ray computerized tomography. Phys Med Biol 1976;21:733–744.

26.

Neuhaus

Große Hokamp

Abdullayev

, et al. Metal artifact reduction by dual-layer computed tomography using virtual monoenergetic images. Eur J Radiol 2017;93:143–148.

27.

Große Hokamp

Neuhaus

Abdullayev

, et al. Reduction of artifacts caused by orthopedic hardware in the spine in spectral detector CT examinations using virtual monoenergetic image reconstructions and metal-artifact-reduction algorithms. Skeletal Radiol 2018;47:195–201.

28.

Sauter

Kopp

Münzel

, et al. Accuracy of iodine quantification in dual-layer spectral CT: influence of iterative reconstruction, patient habitus and tube parameters. Eur J Radiol 2018;102:83–88.