Introduction

The quantification of the degree of stenosis (DoS) of carotid artery is considered up to now the leading parameter for the choice of the therapeutic option in symptomatic subject [1]. Currently, the North American Symptomatic Carotid Endarterectomy Trial (NASCET) method, originally used in the angiographic analysis of the carotid arteries, is the reference modality for the quantification of DoS in the CT exams [2, 3].

With the introduction of dual energy CT (DECT) scanners, it is possible to generate multiple datasets with different level of kilo-electronvolt (keV) [4]. The advantage of this technology is the possibility to observe the variation of attenuations in terms of Hounsfield units (HU) of a tissue according to the level of energy selected increasing the number of information [5,6,7,8,9].

In particular, iodinated contrast material shows a great variation depending on the energy level; in fact, as the keV gets closer to the k-shell binding energy of the atom of iodine, there is an increase of photoelectric effect with consequent higher HU [10]. In the study of carotid arteries, the exams are performed with an arterial phase with a high iodine concentration in the arterial vessel; therefore, this target is significantly reactive to the keV selected [11]. The attenuation of the opacified blood according to the variation of the keV could determine variation in perception and measurements of the vessels size with subsequent variation in the DoS.

Therefore, the purpose of this study is to assess if variation of keV with DECT scanner significantly affects the DoS.

Material and methods

Study design and patient population

This is a retrospective study. The institutional review board approval for this study was obtained and patients’ consent was waived. All the DECT of carotid arteries performed from March 2014 to June 2015 were included for a total of 53 subjects (37 males; mean age 67 ± 11 years; age range 47–83 years). CT of carotid arteries is performed in case of pathological findings in US exams or for subjects with occurrence of acute cerebrovascular events.

CT technique

All patients were studied with a multi-energy CT scanner (Discovery HD 750, Gemstone Spectral Imaging, GE Healthcare, Milwaukee, Wis USA). Angiographic phase imaging was performed by injecting 60 mL of contrast medium (Ultravist 370; Bayer Schering Pharma AG, Berlin, Germany) through an antecubital vein using a power injector at a flow rate of 5 mL/s and an 18-gauge intravenous catheter. A smart prep technique was used to calculate the correct timing of the CT acquisition, and each patient received a 15-mL timing bolus of contrast medium to synchronize the data acquisition with the arrival of contrast material in the carotid arteries. The angiographic acquisition was performed breath-hold in caudo-cranial direction and included the carotid siphon. The convolution kernel used was intermediate.

Image reconstruction and plaque analysis

Images were reconstructed on a dedicated workstation (Advantage Windows Workstation version 4.4, GE Healthcare) and from the CT raw data multiple datasets were generated at the following monochromatic energy levels: 66, 70, 77, and 86 keV (Fig. 1). Two radiologists (blinded for peer review) independently performed all measurements for quantification of the degree of stenosis and the analysis was performed according to the NASCET technique. Because of the measurement of the degree of stenosis is affected by the window-level and window-width settings, a fixed parameter was used according to Saba et al [2] (window width at 850 HU, level of 300 HU).

Fig. 1
figure 1

Example of plaque at different keV energy levels. In case the left ICA plaque of a 75-year-old subject is visualized at different energy levels: panel a shows 66 keV; panel b shows 70 keV; panel c shows 77 keV; panel d shows 86 keV

Full size image

Statistical analysis

Statistical analysis was performed with the SPSS 13.0 statistical package (SPSS Inc., Chicago, IL). The normality of each continuous variable group was tested using the Kolmogorov-Smirnov Z test. Continuous data were described as the mean value ± SD or as median [Q1−Q3] if not normally distributed. Wilcoxon test was used to test the differences between the HU values in the plaques at different keV; it was also used to check whether a statistically significant difference was present in degree of stenosis according the selected keV. Concordance among the different keV as well as the inter-reader variability was assessed with the Bland-Altman analysis. Graphics were plotted with MedCalc 15.0 software (MedCalc, Mariakerke, Belgium). A p value < 0.05 was considered significant and all values were calculated using a two-tailed significance level.

Results

General results

Patient characteristics are summarized in the Table 1. Of the 53 patients, 36 were referred to radiology to perform CT in case of pathological findings in US exams and 17 were referred due to the occurrence of acute cerebrovascular events. None of the patients were excluded from this study for sub-optimal image quality; therefore, a total of 106 carotid arteries were analyzed. An overall view of the results is given in Table 2. The Wilcoxon analysis showed a statistically significant difference in the DoS assessment among the different keVs selected (Table 3). The Bland-Altman for the different keV analyses showed two findings: the DoS difference had a linear relation with the keV difference (the bigger is the difference in keV, the bigger is the variation in DoS). The second finding was that for different keVs, the difference in DoS is reduced with its increase: therefore, for mild grades of stenosis there was a maximum variability whereas for severe grades of stenosis the variation due to the difference in keV was minimal (Fig. 2). We also tested the inter-observer agreement at the different energy level using and the Bland-Altman plots (Fig. 3) and it observed an optimal agreement between the two readers.

Table 1 Patients characteristics
Full size table
Table 2 Overall view of the findings (in percentage)
Full size table
Table 3 Wilcoxon analysis
Full size table
Fig. 2
figure 2

Bland-Altman plots analysis of the DoS according to the keV difference. Panel a shows the DoS difference between 66 and 70 keV. Panel b shows the DoS difference between 66 and 77 keV. Panel c shows the DoS difference between 66 and 86 keV. Panel d shows the DoS difference between 70 and 77 keV. Panel e shows the DoS difference between 70 and 86 keV. Panel f shows the DoS difference between 77 and 86 keV

Full size image
Fig. 3
figure 3

Inter-observer agreement analysis for the DoS quantification. Panel a shows the DoS difference between at 66 keV between reader 1 and reader 2. Panel b shows the DoS difference between at 70 keV between reader 1 and reader 2. Panel c shows the DoS difference between at 77 keV between reader 1 and reader 2. Panel d shows the DoS difference between at 86 keV between reader 1 and reader 2

Full size image

Discussion

Currently, the quantification of the degree of stenosis is considered a key parameter for the choice of the therapeutic option and one of the most frequently used techniques to quantify DoS of carotid artery is the CT [1].

With the introduction of DECT scanner, it is possible to reconstruct the acquired dataset with different mono-energetic keV level. Nowadays, different dual energy CT technologies are currently available but all these approaches offer the potentiality to have monochromatic images. This approach allows the visualization of the images with different tissue attenuation values according to the specific diagnostic requirement.

DECT implies the acquisition of two datasets utilizing two different X-ray energy spectra [5, 10] with the potential to characterize different materials on the basis of their elemental compositions (atomic number) [10]. DECT material decomposition is based on the photoelectric effect that is highly dependent on the atomic number of the absorber and to the binding energy of the innermost electron shell (the k-shell). The probability of photoelectric absorption increases substantially as the energy of the incident photon approaches the K-shell binding energy, the “K-edge.” Therefore, the simultaneous CT acquisition of two X-ray spectra at different peak tube voltages allows for the detection of specific attenuation characteristics based on differing energy levels for material with high atomic number such as iodine (Z = 53) [6, 7].

In the previous years, some papers have been published covering the application of DECT in definition of carotid artery pathology [6, 7, 12, 13], in particular exploring the variation of the HU attenuation [8, 14], or the change in size of the plaques [7] according to the variation of the energy level used but, to the best of our knowledge this is the first research study covering the relationship between the use of the different keV levels and the degree of stenosis.

Because of different iodine attenuation values depending on the keV selected, we aimed to assess if variation of keVs used during the reconstruction process significantly affects the carotid DoS quantification. Recently published studies [15,16,17,18] demonstrated that in the supra-aortic carotid artery, the image quality is better using low keV. In particular, Leithner et al [17] found that the use of protocols with very low keV (40 keV) showed similar and, in some cases, better image quality when compared with protocols with higher values. However, the effect of variation of keV to DoS remains to be disclosed.

In our study, Wilcoxon analysis showed significant differences in the DoS quantification among the different keVs selected (Table 1). This finding suggests that the different attenuations, due the keV, significantly affect the DoS values.

Moreover, it is important to remember that two of the most recurrent artifacts linked to CT arterial vessel analysis are the edge blur and the halo artifact [19, 20]. The edge blur artifact consists in the presence of a blurred edge at the level of the outer luminal margin of the vessel. The halo artifact refers to peri-luminal increased attenuation; both these artifacts are assessed on a scale as a percentage relative to the luminal diameter. These two conditions are affected by the entity of attenuation of the opacified blood in the vessel lumen and may be the cause differences in the carotid DoS quantified through different keV levels.

However, the effect due to the different attenuation values dependent on the keV selected and the consequent different entity edge blur and the halo artifacts could in part explain the variation in the DoS. The second effect could be due to the tissue composition of the plaque. In this study, we did not analyze the impact of the different types of plaque (calcified, mixed, fatty) to the DoS variation because this type of analysis would require significant bigger cohort size due to the need to compare three different groups. However, it is important to underline that the different plaque composition could play a significant role as demonstrated in previously published papers: Mannelli et al [14] found that there is an overestimation of calcified stenosis using low keV values for image reconstruction and similar findings were found in other body regions such as the coronary arteries and lower extremity [21, 22].

Other findings were also detected with the Bland-Altman analysis. First, the magnitude of the difference in terms of DoS increases with the keV difference (the bigger is the difference in keV between the two reconstructions, the bigger is the variation in DoS). The second effect showed by the Bland-Altman analysis is that the difference is reduced with the increase of the DoS: therefore, for a moderate level of stenosis, the variability is the highest whereas for severe grades of stenosis the variation due to the difference in keV is minimal. The increased variability for moderate level of stenosis could represent an important clinical point. This is because this category of patients (moderate level of stenosis) could be treated with different approaches (revascularization, best medical treatment) and the correct quantification of the degree of stenosis currently represents the key parameter of this choice. Moreover, this finding implies, as simple physical consequence, that also with the use of a conventional CT scanner could be present in the same subjects a variability depending on the level of energy applied. This is because the higher or lower use of kV (non-monocromatic) determines an increased or reduced photoelectric effect.

In this study, there are some limitations. The first one is related to the small cohort of subjects we analyzed (n = 53) and the heterogeneity in the validation process with ultrasound; dual-modality validation could be a solution to this weakness. The second one is related to the fact that we focused our attention as the direct consequence of the keV to the opacified level, but also the type of the plaque could play a role: based on CT imaging physics, this is a well-known phenomenon; the degree of calcium blooming is increased on the lower keV monochromatic images. Therefore, the degree of stenosis of a heavily calcified carotid plaque can be overestimated on low keV images, and by the same token, underestimated on the high keV images. We did not categorize the effect of the keV according to the type of plaque (fatty, mixed, calcified) because of the small number of subjects that do not allow to perform the statistical analysis regarding this further interaction.

Conclusion

In conclusion, we found that keV value significantly affects the quantification of the DoS of the carotid artery; therefore, a standardization in the energy levels used for the image reconstruction could potentially be considered.