Accuracy of virtual non-contrast images from dual-energy integrating detector CT and photon-counting detector CT at high iodine concentrations: a head phantom study

Accuracy of virtual non-contrast images from dual-energy integrating detector CT and photon-counting detector CT at high iodine concentrations: a head phantom study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Accuracy of virtual non-contrast images from dual-energy integrating detector CT and photon-counting detector CT at high iodine concentrations: a head phantom study Risto Grkovski, Zsolt Kulcsar, Sebastian Winklhofer This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7345438/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Objectives To evaluate the accuracy of virtual non-contrast (VNC) images at multiple radiation doses and high iodine concentrations using a head CT phantom with dual-energy integrating detector CT (EID-CT; TwinSpiral DECT) and photon-counting detector CT (PCD-CT). Materials and Methods An anthropomorphic head phantom containing brain tissue inserts and varying iodine concentrations (43.75, 175 and 350 mg/ml) was scanned three times with EID-CT and PCD-CT at 100%, 80%, 75%, 65%, 50%, and 25% radiation dose from which standard (mixed) and VNC images were obtained. Attenuations (HU) were measured in standard CT and corresponding VNC images, and both absolute and signed errors of VNC were calculated. For statistical analysis, data were reshaped from wide to long format. Two-way ANOVAs were conducted, considering signed and absolute errors with radiation dose and iodine concentration as factors. Post hoc Tukey tests were applied if p < 0.05. Normality and homoscedasticity were checked via residual diagnostics. If assumptions were violated, nonparametric methods were used. Additionally, HU of fat, bone, and white matter from EID-CT at 100% dose were compared with 16 patient scans to confirm phantom realism. Results ANOVA showed radiation dose significantly affected VNC errors in both scanners, though only 25% vs. 100% dose in PCD-CT was significant (p = 0.0383) after post-hoc Tukey; in EID-CT, no pairwise dose differences were significant (p ≥ 0.07). Iodine concentration exerted a stronger influence: in EID-CT, 350 mg/ml differed from 0, 43.75, and 175 mg/ml (p < 0.001), producing errors up to + 13.7 ± 1.1 HU. In PCD-CT, 175 mg/ml and 350 mg/ml differed from 0 mg/ml (p < 0.05) and from 43.75 mg/ml (p < 0.001), showing a negative bias up to − 10.6 ± 1.5 HU. While dose reductions had limited impact on overall VNC accuracy, higher iodine concentrations (175, 350 mg/ml) caused significant errors in both scanners, albeit with opposite signed biases. Conclusions High iodine concentrations caused significant VNC errors in both scanners—EID-CT overestimating and PCD-CT underestimating the standard baseline—yet both provided substantial visual contrast removal, with radiation dose reductions only rarely impacting accuracy. Dual-energy computed tomography photon counting computed tomography virtual non contrast images iodine maps overestimation underestimation low-dose imaging Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 INTRODUCTION Computed tomography (CT) with spectral information can be achieved using different methods. Several dual-energy techniques (DECT) are employed with energy-integrating detector CT (EID-CT), including dual-source scanning, single-source scanning with kV switching, single-source scanning with split-filter, or single-source with a dual-layer detector. In contrast, photon-counting detector CT (PCD-CT) enables spectral resolution at the detector level by measuring the energy of individual photons[ 1 – 3 ]. In neuroradiological imaging, dual-energy EID-CT minimizes image artifacts from various sources, aids in the differentiation iodine deposition from hemorrhage to guide patient management post-thrombectomy and tumor diagnosis, predicts hematoma expansion and traumatic brain injury outcomes through iodine quantification, and enhances visualization of extra-axial lesions and adjacent vessels by distinguishing hemorrhage from brain parenchymal mineralization using calcium and bone subtraction [ 4 ]. Furthermore, a recent review of initial clinical results of PCD-CT [ 5 ] summarized, that PCD-CT, though in its infancy, has shown potential in reducing dosage, suppressing noise and artifacts, enhancing spatial resolution, and expanding spectral information use, which could revolutionize neuroimaging and improve clinical applications such as temporal bone imaging, stent visualization, head and neck angiographies, and stroke imaging. Virtual non-contrast (VNC) images are based on a post-processing technique that uses iodine subtraction to simulate true non-contrast images of contrast-enhanced CT scans in DECT [ 6 – 8 ]. However, they have demonstrated varying levels of accuracy in estimating attenuation values of the true non-contrast (TNC) images across different spectral CT technologies. Studies have shown that VNC images often closely approximate the TNC images, particularly in soft tissues, with minimal differences in attenuation values [ 9 – 18 ]. Nevertheless, discrepancies that emerge in tissues with higher iodine concentrations or more complicated compositions (e.g., spongy bone) can pose a challenge—particularly when additional confounding materials, such as parenchymal calcifications, act as a “fourth material,” introducing beam-hardening or saturation effects that hamper VNC accuracy [ 18 – 23 ]. Larger patient size can further affect inaccuracy of VNC images, leading to greater variabilities of estimations [ 23 ]. In one phantom study, VNC images for contrast materials and simulated soft tissues were found to be fairly accurate, with the exception of synthetic bone [ 24 ]. Despite these hurdles, improvements in photon-counting CT and dual-energy spectral CT have bolstered overall accuracy, keeping VNC errors within clinically acceptable margins [ 25 , 26 ]. Nonetheless, given that the accuracy can be influenced by body habitus, different iodine concentrations, and scanner technology, clinicians should exercise caution when interpreting VNC data in routine practice. The aim of our study was to evaluate the accuracy of virtual non-contrast (VNC) images at different radiation doses and elevated iodine contrast concentrations than previously tested in a head phantom, using third generation spectral CT systems: dual-energy-integrating detector CT (TwinSpiral DECT) and photon-counting detector CT. MATERIALS AND METHODS Phantom An anthropomorphic acrylic CT phantom representing a human head was constructed, similar to those as in other studies [ 27 – 29 ]. The phantom consisted of three cylindrical segments (Fig. 1 ), each measuring 16 cm in diameter and 6 cm in height. The central cylinder contained inserts encompassing its whole height in which material simulating brain tissue and iodine solutions of different concentrations were placed. The inserts had a diameter of 1 cm and were positioned in the center and two on each side, perpendicular to each other, thus nine inserts in total. The inserts were filled with 0.9% NaCl solution, calcium carbonate (CaCO 3 – simulating bony tissue, [ 30 – 32 ]), butter (simulating fatty tissue, [ 33 ]), agar (simulating brain tissue, [ 34 , 35 ]), and iodine contrast media of different concentrations (100% contrast solution (Xenetix® 350 mg/ml; Guerbet, Aulnay-Sous-Bois, France)), 50% contrast solution (175 mg/ml), 25% contrast solution (87,5 mg/ml) and 12.5% contrast solution (43,75 mg/ml). CT Imaging Imaging was performed on two CT scanners with different underlying technologies: a single-source energy-integrating detector CT (EID-CT; Somatom X.cite, Siemens Healthcare, Erlangen, Germany) in TwinSpiral dual-energy mode [ 36 , 37 ] and a dual-source PCD-CT equipped with two cadmium telluride PCDs (PCD-CT; Siemens NAEOTOM Alpha; Siemens Healthineers, Erlangen, Germany) [ 25 , 26 , 38 ]. For each scanner, six image acquisition protocols were applied, representing varying radiation dose levels commonly used in clinical practice. From each acquisition, standard (mixed) images and virtual non-contrast (VNC) images were reconstructed for subsequent analysis. Table 1 Acquisition and reconstruction parameters used in the TwinSpiral DECT and PCD-CT: Parameter TwinSpiral DECT (EID-CT) PCD-CT Rotation Time 1 s 0.5 s Pitch 0.55 0.55 Slice Acquisition 64 × 0.6 mm 144 × 0.4 mm Tube Current-Time Products (mAs)* 220, 176, 165, 143, 110, 55 249, 199, 187, 162, 125, 62 Protocol’s CTDIvol (mGy)* 43.7, 35.0, 32.8, 28.4, 21.9, 10.9 42.0, 33.6, 31.5, 27.3, 21.0, 10.5 Matrix Size 512 × 512 pixels 512 × 512 pixels Imaging Mode Dual-energy mode at 80/Sn150 kVp (tin filter for the second tube) Spectral mode (QuantumPlus) at 120 kVp Reconstruction Kernel Quantitative Qr40 Quantitative Qr40 Slice Thickness 4 mm 0.6 mm Increment 4 mm 0.6 mm * Tube current-time products and protocol’s CTDIvol values correspond to 100, 80, 75, 65, 50, and 25% of the standard clinical dose respectively. For each scanner, each dose setting was scanned three times, resulting in a total of 18 scans per scanner (6 settings × 3 repetitions) and 36 scans overall (18 × 2 scanners). Patients Head CT scans were conducted on a Somatom X.cite single-source CT scanner from Siemens Healthineers, Erlangen, Germany. The scans were performed in TwinSpiral dual-energy mode with tube voltages of 80 and 150 kVp, the latter of which utilized tin filtration. The quality reference tube current-time products were 220 and 179 mAs, respectively, and were modulated using CAREDose4D. The slice acquisition was 64 × 0.6 mm, the rotation time was 1.0 s, and the pitch was 0.55. The protocol’s mean volume CT dose index (CTDIvol) was 43.7 ± 3.4 mGy. Non-contrast brain CT images from 16 patients, scanned with TwinSpiral DECT during follow-up after mechanical thrombectomy due to ischemic stroke using the above protocol, were retrospectively analyzed, HU measurements were made in the area of healthy cerebral white matter tissue, subcutaneous fat and scalp bone, and matched with standardized HU values of these tissues and phantom inserts imitating bone, brain white matter and fat. These measurements represented the values at 100% of standard dose used in clinical practice by the scanner. Only patients with no visible brain herniation or any lesions causing intracranial mass effect were included. All patients were older than 18 years. For PCD-CT no prior brain scans on patients were available. All procedures were performed in accordance with local and federal regulations and the Declaration of Helsinki. The study was approved by the local ethics committee (approval number blinded for the review purposes) and patient consent was obtained from each participant. Quantitative Analysis In the phantom datasets, regions of interest (ROIs) were placed in the inserts containing the different materials. ROIs were drawn as large as possible (diameter ≥ 6 mm) (Fig. 2 ), but not too close to the insert’s edges. Placement of ROIs was performed on standard (mixed) CT images and VNC images, and were then copied to the corresponding position on the other two scans. From the three ROIs of a measurement, the mean CT attenuation (HU) was recorded and an average of each three scans for each measurement was considered representative for further analyses. All measurements were repeated four times to ensure consistency. Based on the acquired parameters, an absolute and signed error vas calculated as follows: VNC error = |(HU VNC – HU base )| Where the HU VNC represents the attenuation on VNC images and HU base corresponds to the attenuation of a standard CT (true non-contrast (TNC)) images in a phantom, respectively. () represents signed and || absolute error. Due to the occurrence of beam-hardening artifacts in some of the images, the insert with 87.5 mg/ml iodine contrast concentration was excluded from further analysis. In the patient scans, ROIs were placed in the white brain matter, cortical bone of the skull and subcutaneous fat (Fig. 3 ). Likewise, a placement of ROI was performed on respectively represented tissue inserts on standard CT and VNC images of the phantom. Statistical Analysis Data were first reshaped from wide to long format to capture replicate measurements. Two-way analyses of variance (ANOVAs) were then conducted on both signed errors (VNC − TNC) and absolute errors |VNC − TNC|, with radiation dose and iodine concentration as fixed factors. Whenever a main effect or interaction had a p-value < 0.05, post hoc Tukey tests were performed to identify specific group differences. Residual diagnostics (QQ plots, histograms, and residuals-versus-fitted plots) were inspected to confirm that parametric assumptions (normality and homoscedasticity) were not severely violated. In addition, Shapiro–Wilk and Kolmogorov–Smirnov tests were used to formally test for normality of residuals. If these diagnostics suggested marked deviations from normality or large outliers, nonparametric methods (for example, Wilcoxon or other rank-based test) were considered as a backup; otherwise, the standard ANOVA approach with post hoc Tukey comparisons was retained. A two-sided p-value < 0.05 was considered statistically significant (95% confidence interval). A Wilcoxon rank-sum was used to compare HU measurements from patients with phantom. All statistical analyses were performed in the Python programming language (version 3.14; Python Software Foundation, Wilmington, DE, USA, https://www.python.org/ ). RESULTS Phantom scans Effect of Radiation Dose In TwinSpiral DECT (EID-CT) (Fig. 4 ), an ANOVA on absolute VNC errors showed a significant main effect of radiation dose (p < 0.001), yet no pairwise dose comparisons were significant after Tukey adjustment (p = 0.07–1.00). Hence, none of the specific radiation dose levels (25%, 50%, 65%, 75%, 80%, 100%) differed significantly from each other in absolute errors, which ranged from 0.7 ± 0.5 HU (at 65% dose) up to 2.4 ± 2.2 HU (at 25% dose) when iodine concentration was 0 mg/ml, and from 8.2 ± 1.1 HU (at 50% dose) up to 13.7 ± 1.1 HU (at 75% dose) at 350 mg/ml. A similar ANOVA on signed errors also indicated a significant overall dose effect (p < 0.001), but again no post hoc pairwise differences reached significance (p ≥ 0.05). In contrast, the PCD-CT’s ANOVA (Fig. 5 ) on absolute errors again showed a significant main effect of dose (p < 0.001), but no individual dose-level comparisons were significant (p ≥ 0.05). Across 0 mg/ml and low iodine concentrations (0 mg/ml or 43.75 mg/ml), absolute errors ranged from 1.6 ± 0.9 HU (at 50% dose) up to 4.9 ± 2.9 HU (at 100% dose). At 350 mg/ml, values spanned 6.8 ± 1.5 HU (at 80% dose) up to 10.6 ± 1.5 HU (at 25% dose). However, a separate signed-error analysis for PCD-CT revealed a specific difference between 25% and 100% radiation dose (p = 0.0383), while other dose-level comparisons remained non-significant (p > 0.05). Effect of Iodine Concentration In TwinSpiral DECT (EID-CT) (Fig. 4 ), iodine concentration strongly influenced the absolute errors (p < 0.001). For instance, 350 mg/ml differed significantly from 0 mg/ml (p < 0.001), from 43.75 mg/ml (p < 0.001), and from 175 mg/ml (p < 0.001). Similarly, 175 mg/ml differed from both 0 mg/ml (p < 0.05) and 43.75 mg/ml (p < 0.001). Consequently, the absolute errors at 0 mg/ml ranged from 0.7 ± 0.5 HU up to 2.4 ± 2.2 HU, whereas at 350 mg/ml they reached 13.7 ± 1.1 HU (75% dose). A similar pattern emerged for signed errors, which confirmed overestimations at higher concentrations: for example, + 5.0 ± 0.7 HU at 175 mg/ml (100% dose) and as high as + 13.7 ± 1.1 HU at 350 mg/ml (75% dose). Similarly, the PCD-CT scanner (Fig. 5 ) likewise exhibited a strong effect of iodine concentration, with absolute errors at 175 mg/ml and 350 mg/ml each differing significantly from 0 mg/ml (p < 0.05). Moreover, for signed errors, 175 mg/ml differed significantly from 43.75 mg/ml (p < 0.001), and 350 mg/ml differed significantly from 43.75 mg/ml (p < 0.001). By contrast, the difference between 175 mg/ml and 350 mg/ml was not significant (p = 0.19). Also, for both signed and absolute errors, 0 mg/ml vs. 43.75 mg/ml was not significant (p = 0.058 for signed errors, p = 0.785 for absolute errors). Across these lower iodine concentrations (0 mg/ml, 43.75 mg/ml), mean absolute errors ranged from 1.6 ± 0.9 HU (50% dose) to 4.9 ± 2.9 HU (100% dose). At 350 mg/ml, values spanned 6.8 ± 1.5 HU (80% dose) to 10.6 ± 1.5 HU (25% dose). A signed-error analysis further confirmed underestimation at higher concentrations (175 mg/ml, 350 mg/ml), which ranged from roughly − 2.7 ± 2.5 HU (100% dose) to − 10.6 ± 1.5 HU (25% dose). Patient scans A total of 16 patients (7 women, 44%, 9 men, 56%, mean age 65 ± 9 years, range 51–78 years) were included in the study analysis. A Wilcoxon rank-sum test showed no statistical difference (p ≥ 0.05) between average values of HU of bone or white matter between the patient and phantom images with TwinSpiral DECT (a dual-energy EID-CT), but did show significant difference of fat tissue (p < 0.05). DISCUSSION In this study, we quantified virtual non-contrast (VNC) measurement errors in two CT scanner systems—TwinSpiral DECT (an EID-CT scanner) and photon counting detector CT (PCD-CT scanner)—across varying radiation dose levels and higher than typical iodine concentrations, simulating contrast extravasation during interventional procedures. Distinguishing blood from iodine contrast with DECT after mechanical thrombectomy allows for more precise identification of potentially complicated bleeding versus contrast extravasation. This differentiation can improve diagnostic confidence in confirming hemorrhage, better delineate the extent of blood extravasation (hematoma) separately from contrast leakage, and ultimately aid in more targeted and effective clinical management, such as surgical or interventional planning, by clearly visualizing the source and nature of the leak. We evaluated two types of error metrics: (1) absolute error, which measured how far a VNC reading deviated from a true reference value in terms of pure magnitude, and (2) signed error, which indicated whether the VNC estimate overestimated (positive sign) or underestimated (negative sign) the true value. Interpreting both metrics together was particularly valuable: absolute error captured the extent of the discrepancy, whereas signed error clarified the direction of that discrepancy. Overall, the most pronounced finding was that iodine concentration strongly influenced the magnitude of VNC errors in both scanners, overshadowing many of the dose-related differences. In the TwinSpiral DECT, absolute errors were smallest (on the order of 1–2 HU) when little or no iodine was present, but rose substantially (up to 13–14 HU) as iodine concentration increased to 350 mg/ml. Likewise, in PCD-CT, errors were low (2–4 HU) at lower iodine levels and climbed to roughly 7–10 HU at higher concentrations. This pattern underscores that as more iodinated contrast remains in the image, the task of “subtracting” it to create a virtual non-contrast image becomes more challenging for both scanning systems. Previous studies similarly showed VNC accuracy being linked to iodine contrast concentration. Harsaker et al. [ 21 ] found errors increased markedly above 50 mg/ml, escalating at 100 mg/ml. Sartoretti et al. [ 25 ] noted a significant drop in accuracy of VNC images in EID-CT beyond 3 mg/ml, whereas in PCD-CT VNC images stayed accurate to up to 5 mg/ml. Holz et al. [ 23 ] reported largest deviations in VNC accuracy at 5.0 mg/ml, and Mergen et al. [ 26 ] observed stable accuracy up to 5 mg/ml. Except for Harsaker et al., however, these studies used concentrations far lower than ours, highlighting the uniqueness of our higher-concentration findings. Furthermore, it is important to note that in our study, despite the inaccuracies of VNC images at higher iodine concentrations (175 and 350 mg/ml), the visual appearance (qualitative difference) of VNC images still reflected a marked reduction in the HU compared to TNC images (Fig. 2 ), which may remain diagnostically useful, even if quantitative accuracy is compromised, as it might occur in vessel perforation during interventional procedure. Interestingly, although the broad trend of escalating errors with higher iodine was similar in both scanners, the direction of the signed error differed. TwinSpiral DECT tended to overestimate the reference value at high iodine levels (i.e., a positive bias of roughly + 5 HU to + 14 HU), whereas PCD-CT consistently showed negative signed errors—an underestimation of the reference—ranging from around − 3 HU to − 10 HU. These opposite directions of biases likely stem from differences in each scanner’s underlying spectral decomposition algorithms, handling extreme scenarios with high iodine concentrations than usually found in soft tissue lesions, or the difference in slice thickness. Holz et al. [ 23 ] similarly found underestimation of VNC images at 0.3 mg/ml iodine contrast concentrations, but then the directions shifted to overestimation at 5.0 mg/ml. Likewise Hua et al. [ 39 ] noted a shift from under- to overestimation of VNC images between 2.5 mg/ml and 5 mg/ml iodine contrast. However, these studies again used lower concentrations, underscoring the novelty of our data. Regarding radiation dose, both scanners showed statistically significant main effects when analyzing absolute and signed errors across different radiation dose levels (25%, 50%, 65%, 75%, 80%, and 100% of the full dose). However, post hoc comparisons rarely produced significant pairwise distinctions. In TwinSpiral DECT, neither the absolute error nor the signed error analyses singled out any one dose level as clearly better or worse once multiple-comparison corrections were applied. Meanwhile, in PCD-CT, a single notable result emerged: at 25% vs. 100% dose, the signed error showed a statistically significant difference (p = 0.0383). Despite these findings, the overall magnitude of errors at different dose levels remained broadly comparable, and any shifts in bias were modest, although an overall slight overestimation has occurred in TwinSpiral DECT and underestimation in PCD-CT. One possible explanation is that while lower-dose scans typically exhibit increased image noise, this additional noise did not manifest as a large difference in HU measurements and thus the average VNC errors—at least not enough to become statistically distinct in most pairwise comparisons. This result suggests that moderate or even substantial dose reductions could be pursued without dramatically compromising VNC accuracy. However, the thinner slice thickness in PCD-CT compared to TwinSpiral DECT might have contributed to more noise due to greater variability of measured HU, which might have caused the significant difference between 25% and 100% radiation dose in PCD-CT. As such, careful consideration should be made when imaging with spectral CT scanner at very low radiation doses. Consistent with our findings, Li et al. [ 24 ] reported significant inaccuracies at 75% and 50% of the clinical dose, whereas Holz et al. [ 23 ] found no significant differences among 10, 15, and 20 mGy acquisitions. Sartoretti et al. [ 25 ] similarly concluded that dose reductions (5, 10, 15 mGy) did not significantly affect VNC accuracy in EID-CT or PCD-CT. A Wilcoxon rank-sum test showed a significant difference (p < 0.05) in the average HU values of fat tissue when comparing images from TwinSpiral DECT with patient data. However, due to hygienic constraints and hospital policy, we were not permitted to use real animal subcutaneous fat to represent human fat tissue. Instead, we used butter, as documented in previous studies [ 33 ], although it does have lower HU values than actual subcutaneous fat. Taken together, the current data shows that iodine concentration is the dominant factor in determining how accurately either scanner can remove or “subtract” iodine in VNC reconstructions. At lower iodine contrast concentrations, both TwinSpiral DECT and PCD-CT produced acceptably small absolute errors, but as iodine concentrations reached 175 mg/ml and above, significantly larger discrepancies from the true non-contrast baseline occurred—albeit still with a clear visual reduction in overall attenuation values. Moreover, the scanners did not err in the same direction: TwinSpiral DECT generally overshot (positive bias), while PCD-CT undershot (negative bias). Meanwhile, although the dose setting exerted a statistically significant effect in a global sense—likely reflecting differences in signal-to-noise ratios and reconstruction constraints—practical pairwise differences rarely reached significance, implying that large dose reductions could often be implemented without incurring a dramatic rise in error levels. Nonetheless, a careful balance is required: if clinicians are monitoring subtle differences in HU (e.g., a few HU changes in a small lesion), lower-dose protocols might introduce noise-related variability or small systematic biases that become clinically important. Limitations There were several limitations to our study. First, for PCD-CT scanner we did not have CT head and brain images from actual patients, since at that time, the patients with neuroradiology related pathologies were not routinely scanned on PCD-CT scanner Second, the phantom used in our experiments did not incorporate manufacturer-produced tissue inserts for real-world correlation. Instead, we relied on alternative materials cited in previous research. Third, the qualitative assessment of attenuation drop on VNC images at high iodine concentrations was not conducted by a trained neuroradiologist. While the differences were so pronounced, that even a non-specialist could readily appreciate them, reducing the need for formal expert confirmation in this particular instance, we acknowledge that a formal evaluation by a neuroradiologist could have further strengthened our findings. Lastly, each scanner was used with a customized slice thickness optimized for its design. As such, instead of making a direct comparison, the results were reported separately to showcase the performance of each scanner under its specific imaging parameters. CONCLUSION AND FURTHER DIRECTIVES In conclusion, our analysis showed that iodine concentration, rather than radiation dose, affected VNC accuracy in both TwinSpiral DECT (EID-CT) and PCD-CT, with higher concentrations leading to significant measurement deviations. Although dose reductions did not significantly degrade performance, TwinSpiral DECT tended to overestimate whereas PCD-CT tended to underestimate HU values at high iodine levels. Refining spectral decomposition algorithms and validating these findings in larger clinical cohorts could further mitigate quantitative biases and strengthen clinical utility. Declarations Conflict of interest SW: Nothing to declare. RG: Nothing to declare. ZK: Nothing to declare References McCollough CH, et al. Dual- and multi-energy CT: principles, technical approaches, and clinical applications. Radiology . 2015;276(3):637-653. doi:10.1148/radiol.2015142631. So A, Nicolaou S. Spectral computed tomography: fundamental principles and recent developments. Korean J Radiol . 2021;22(1):86-96. doi:10.3348/kjr.2020.0144. Lennartz S, et al. Virtual unenhanced images: qualitative and quantitative comparison between different dual-energy CT scanners in a patient and phantom study. Invest Radiol . 2022;57(1):52-61. doi:10.1097/RLI.0000000000000802. Gibney B, et al. A review of the applications of dual-energy CT in acute neuroimaging. Can Assoc Radiol J . 2020;71(3):253-265. doi:10.1177/0846537120904347. Abel F, Schubert T, Winklhofer S. Advanced neuroimaging with photon-counting detector CT. Invest Radiol . 2023;58(7):472-481. doi:10.1097/RLI.0000000000000984. Sauter AP, Muenzel D, Dangelmaier J, et al. Dual-layer spectral computed tomography: virtual non-contrast in comparison with true non-contrast images. Eur J Radiol . 2018;104:108-114. doi:10.1016/j.ejrad.2018.05.007. Ananthakrishnan L, Rajiah P, Ahn R, et al. Spectral detector CT-derived virtual non-contrast images: comparison of attenuation values with unenhanced CT. Abdom Radiol (NY) . 2017;42(3):702-709. doi:10.1007/s00261-016-1036-9. Borges AP, Antunes C, Curvo-Semedo L. Pros and cons of dual-energy CT systems: “one does not fit all”. Tomography . 2023;9(1):195-216. doi:10.3390/tomography9010017. Ananthakrishnan L, Rajiah P, Ahn R, et al. Spectral detector CT-derived virtual non-contrast images: comparison of attenuation values with unenhanced CT. Abdom Radiol (NY) . 2017;42(3):702-709. doi:10.1007/s00261-016-1036-9. Javadi S, Elsherif S, Bhosale P, et al. Quantitative attenuation accuracy of virtual non-enhanced imaging compared to true non-enhanced imaging on dual-source dual-energy CT. Abdom Radiol (NY) . 2020;45(4):1100-1109. doi:10.1007/s00261-020-02415-8. Mingkwansook V, Puwametwongsa K, Watcharakorn A, Dechasasawat T. Comparative study of true and virtual non-contrast imaging generated from dual-layer spectral CT in patients with upper aerodigestive tract cancer. Pol J Radiol . 2022;87:e678-e687. doi:10.5114/pjr.2022.123829. Meyer M, Nelson RC, Vernuccio F, et al. Virtual unenhanced images at dual-energy CT: influence on renal lesion characterization. Radiology . 2019;291(2):381-390. doi:10.1148/radiol.2019181100. Zhang X, Zhang G, Xu L, et al. Utilisation of virtual non-contrast and virtual mono-energetic images from dual-layer spectral CT for renal cell carcinoma: image quality and radiation dose. Insights Imaging . 2022;13(1):12. doi:10.1186/s13244-021-01146-8. Takane Y, Sato K, Kageyama R, Takano H, Kayano S. Accuracy of virtual non-contrast images with different algorithms in dual-energy computed tomography. Radiol Phys Technol . 2022;15(3):234-244. doi:10.1007/s12194-022-00668-0. Niehoff JH, Woeltjen MM, Laukamp KR, Borggrefe J, Kroeger JR. Virtual non-contrast versus true non-contrast computed tomography: initial experiences with a photon-counting scanner approved for clinical use. Diagnostics (Basel) . 2021;11(12):2377. doi:10.3390/diagnostics11122377. Verstraeten S, Ansems J, van Ommen W, van Linden F, Looijmans F, Tesselaar E. Comparison of true non-contrast and virtual non-contrast images for renal lesion characterization with detector-based spectral CT. Br J Radiol . 2023;96(1149):20220157. doi:10.1259/bjr.20220157. Wen D, Pu Q, Peng P, et al. Comparison of virtual and true non-contrast images from dual-layer spectral detector CT in patients with colorectal cancer. Quant Imaging Med Surg . 2024;14(9):6260-6272. doi:10.21037/qims-24-535. Toepker M, Moritz T, Krauss B, et al. Virtual non-contrast in second-generation dual-energy CT: reliability of attenuation values. Eur J Radiol . 2012;81(3):e398-e405. doi:10.1016/j.ejrad.2011.12.011. Lehti L, Söderberg M, Höglund P, Nyman U, Gottsäter A, Wassélius J. Reliability of virtual non-contrast CT angiography: comparing it with the real deal. Acta Radiol Open . 2018;7(7-8):2058460118790115. doi:10.1177/2058460118790115. Dinkel J, Khalilzadeh O, Phan CM, et al. Technical limitations of dual-energy CT in neuroradiology: 30-month institutional experience and literature review. J Neurointerv Surg . 2015;7(8):596-602. doi:10.1136/neurintsurg-2014-011241. Harsaker V, Jensen K, Andersen HK, Martinsen AC. Quantitative benchmarking of iodine imaging for two CT spectral imaging technologies: a phantom study. Eur Radiol Exp . 2021;5(1):24. doi:10.1186/s41747-021-00224-2. Kanatani R, Shirasaka T, Kojima T, Kato T, Kawakubo M. Influence of beam hardening in dual-energy CT imaging: phantom study for iodine mapping, virtual monoenergetic imaging and virtual non-contrast imaging. Eur Radiol Exp . 2021;5(1):18. doi:10.1186/s41747-021-00217-1. Holz JA, Alkadhi H, Laukamp KR, et al. Quantitative accuracy of virtual non-contrast images from spectral detector CT: an abdominal phantom study. Sci Rep . 2020;10(1):21575. doi:10.1038/s41598-020-78518-5. Li B, Pomerleau M, Gupta A, Soto JA, Anderson SW. Accuracy of dual-energy CT virtual unenhanced and material-specific images: a phantom study. AJR Am J Roentgenol . 2020;215(5):1146-1154. doi:10.2214/AJR.19.22372. Sartoretti T, Mergen V, Higashigaito K, Eberhard M, Alkadhi H, Euler A. Virtual non-contrast imaging of the liver with photon-counting detector CT: a systematic phantom and patient study. Invest Radiol . 2022;57(7):488-493. doi:10.1097/RLI.0000000000000860. Mergen V, Racine D, Jungblut L, et al. Virtual non-contrast abdominal imaging with photon-counting detector CT. Radiology . 2022;305(1):107-115. doi:10.1148/radiol.213260. Chou R, Li JH, Ying LK, Lin CH, Leung W. Quantitative assessment of three vendors’ metal artifact reduction techniques for CT imaging using a customized phantom. Comput Assist Surg . 2019;24(suppl 2):34-42. doi:10.1080/24699322.2019.1649075. Feng M, Ji X, Zhang R, Treb K, Dingle AM, Li K. An experimental method to correct low-frequency concentric artifacts in photon-counting CT. Phys Med Biol . 2021;66(17):10808/1361-6560/ac1833. doi:10.1088/1361-6560/ac1833. Nomura Y, Watanabe H, Tomisato H, Kawashima S, Miura M. Gumbel distribution-based technique enables quantitative comparison between streak metal artifacts of multidetector-row CT and cone-beam CT: a phantom study. Phys Eng Sci Med . 2023;46(2):801-812. doi:10.1007/s13246-023-01252-5. Badiuk SR, Sasaki DK, Rickey DW. An anthropomorphic maxillofacial phantom using three-dimensional printing, polyurethane rubber and epoxy resin for dental imaging and dosimetry. Dentomaxillofac Radiol . 2022;51(1):20200323. doi:10.1259/dmfr.20200323. Cho HM, Ding H, Barber WC, Iwanczyk JS, Molloi S. Microcalcification detectability using a bench-top prototype photon-counting breast CT based on a silicon strip detector. Med Phys . 2015;42(7):4401-4410. doi:10.1118/1.4922680. Cao Q, Sisniega A, Stayman JW, Yorkston J, Siewerdsen JH, Zbijewski W. Quantitative cone-beam CT of bone mineral density using model-based reconstruction. Proc SPIE Int Soc Opt Eng . 2019;10948:109480Y. doi:10.1117/12.2513216. Mostafavi MR, Ernst RD, Saltzman B. Accurate determination of chemical composition of urinary calculi by spiral computerized tomography. J Urol . 1998;159(3):673-675. Menikou G, Dadakova T, Pavlina M, Bock M, Damianou C. MRI-compatible head phantom for ultrasound surgery. Ultrasonics . 2015;57:144-152. doi:10.1016/j.ultras.2014.11.004. Krüger MT, Coenen VA, Egger K, Shah M, Reinacher PC. Development of a standardized cranial phantom for training and optimization of functional stereotactic operations. Stereotact Funct Neurosurg . 2018;96(3):190-196. doi:10.1159/000489581. Grkovski R, Acu L, Ahmadli U, et al. Dual-energy CT in stroke imaging: value of a new acquisition technique for ischemia detection after mechanical thrombectomy. Clin Neuroradiol . 2023;33(3):747-754. doi:10.1007/s00062-023-01270-6. Grkovski R, Acu L, Ahmadli U, et al. A novel dual-energy CT method for differentiating intracerebral haemorrhage from contrast extravasation after endovascular thrombectomy: feasibility and first results. Clin Neuroradiol . 2023;33(1):171-177. doi:10.1007/s00062-022-01198-3. Racine D, Mergen V, Viry A, et al. Photon-counting detector CT with quantum iterative reconstruction: impact on liver lesion detection and radiation dose reduction. Invest Radiol . 2023;58(4):245-252. doi:10.1097/RLI.0000000000000925. Hua CH, Shapira N, Merchant TE, Klahr P, Yagil Y. Accuracy of electron density, effective atomic number and iodine concentration determination with a dual-layer dual-energy CT system. Med Phys . 2018;45(6):2486-2497. doi:10.1002/mp.12903. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7345438","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":503207971,"identity":"d745807f-4207-4a55-9c6a-05c9a004bd18","order_by":0,"name":"Risto Grkovski","email":"","orcid":"","institution":"Department of Neuroradiology, Clinical Neuroscience Center, University Hospital of Zurich, University of Zurich","correspondingAuthor":false,"prefix":"","firstName":"Risto","middleName":"","lastName":"Grkovski","suffix":""},{"id":503207972,"identity":"a5189b10-02fa-420a-b957-6b482cc1a569","order_by":1,"name":"Zsolt Kulcsar","email":"","orcid":"","institution":"Department of Neuroradiology, Clinical Neuroscience Center, University Hospital of Zurich, University of Zurich","correspondingAuthor":false,"prefix":"","firstName":"Zsolt","middleName":"","lastName":"Kulcsar","suffix":""},{"id":503207974,"identity":"918e0f83-b476-441a-aa3d-9bd7b374855c","order_by":2,"name":"Sebastian Winklhofer","email":"data:image/png;base64,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","orcid":"","institution":"Department of Neuroradiology, Clinical Neuroscience Center, University Hospital of Zurich, University of Zurich","correspondingAuthor":true,"prefix":"","firstName":"Sebastian","middleName":"","lastName":"Winklhofer","suffix":""}],"badges":[],"createdAt":"2025-08-11 10:38:24","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7345438/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7345438/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":89640273,"identity":"7a15cb85-ba26-4783-81fc-45204c6ce424","added_by":"auto","created_at":"2025-08-22 08:04:15","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":184269,"visible":true,"origin":"","legend":"\u003cp\u003eSchematic (A), pictorial (B) and image acquisition (C) representation of the anthropomorphic acrylic CT phantom: Image A: NaCl – 0.9% NaCl solution, Fat – butter, Ca - calcium carbonate (CaCO\u003csub\u003e3\u003c/sub\u003e), Brain – agar. 12.5%, 25%, 50%, 100% – different iodine concentrations diluted with 0.9% NaCl solution (43.75, 87.5, 170, 350 mg/ml respectively). The dark circle represents an insert, which was not used for measurements. The phantom was 16 cm in diameter, with numerous holes 1 cm in diameter for different inserts placement, placed on the imaginary diameter lines perpendicular to each other. Image B showing two additional cylinders, in addition to the one with the inserts, in order to simulate length of the head (18 cm in length) and prevent x-ray from entering into the ends of the holes with inserts. Image C: Phantom placed in the gantry of the photon counting detector CT. The phantom was fixed using towels in order to preserve the correct orientation through different scans.\u0026nbsp;\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7345438/v1/cd890023fe7c13f6b37a9abc.png"},{"id":89640266,"identity":"c8e610d8-cab5-4501-985e-e16bc1a25d36","added_by":"auto","created_at":"2025-08-22 08:04:15","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":110068,"visible":true,"origin":"","legend":"\u003cp\u003eAn example of spectral imaging of the phantom. A – standard (mixed) CT image from TwinSpiral dual-energy CT (DECT), B - VNC image from TwinSpiral DECT, C – standard CT image from photon counting detector CT (PCD-CT), D - VNC image from PCD-CT. A blue circle represents an example of HU measurement of an insert. A notable attenuation reduction can be observed between standard CT images (red arrows) and VNC images (yellow arrows) for both 175 mg/ml (positioned on the right) and 350 mg/ml (positioned on the bottom) iodine contrast inserts.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7345438/v1/2d868d9373f742f12dcc4945.png"},{"id":89642515,"identity":"81d46996-da42-4abc-82cc-3f4469b0e7a5","added_by":"auto","created_at":"2025-08-22 08:20:15","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":132344,"visible":true,"origin":"","legend":"\u003cp\u003eNon-contrast brain CT image of a patient scanned with TwinSpiral dual-energy CT during follow-up after mechanical thrombectomy due to ischemic stroke. A: standard CT image, B: virtual non-contrast (VNC) image. Yellow circle – region of interest (ROI) inside the brain white matter, blue circle – ROI inside the cortical bone, green circle – ROI inside the subcutaneous fatty tissue.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7345438/v1/73ea2e90524d15e1cae6690d.png"},{"id":89640268,"identity":"443298a3-f065-4d7f-8d56-725973a1ec1b","added_by":"auto","created_at":"2025-08-22 08:04:15","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":172984,"visible":true,"origin":"","legend":"\u003cp\u003eThe subgroup analyses of quantitative data in the phantom are depicted in this figure, focusing on VNC error for \u003cstrong\u003eTwinSpiral DECT (EID-CT)\u003c/strong\u003e. The data are presented through notched boxplots, where the bold black lines within the plots indicate the medians. The edges of the boxplots correspond to the 25th and 75th percentiles, while the whiskers extend from these edges to the maximum and minimum values within 1.5 times the interquartile range. Note increased signed and absolute errors at higher iodine concentrations, with a positive sign representing overestimation, while radiation dose having less of an effect of the VNC errors.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7345438/v1/5362781204ca6a314f001b21.png"},{"id":89640274,"identity":"fe798679-9307-45f2-b5fe-a234a55f949e","added_by":"auto","created_at":"2025-08-22 08:04:15","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":157282,"visible":true,"origin":"","legend":"\u003cp\u003eThe subgroup analyses of quantitative data in the phantom are depicted in this figure, focusing on VNC error for \u003cstrong\u003ephoton counting detector CT\u003c/strong\u003e. The data are presented through notched boxplots, where the bold black lines within the plots indicate the medians. The edges of the boxplots correspond to the 25th and 75th percentiles, while the whiskers extend from these edges to the maximum and minimum values within 1.5 times the interquartile range. Note increased signed and absolute errors at higher iodine concentrations, with a negative sign representing underestimation, while radiation dose having less of an effect of the VNC errors.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7345438/v1/c0ac5ee112e45c8212453e35.png"},{"id":98776559,"identity":"a4fda0f1-cffc-4155-a22a-5faa71580f1b","added_by":"auto","created_at":"2025-12-22 12:23:04","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1242743,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7345438/v1/62432d2f-979a-4376-9370-9bfb84fb1842.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Accuracy of virtual non-contrast images from dual-energy integrating detector CT and photon-counting detector CT at high iodine concentrations: a head phantom study","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eComputed tomography (CT) with spectral information can be achieved using different methods. Several dual-energy techniques (DECT) are employed with energy-integrating detector CT (EID-CT), including dual-source scanning, single-source scanning with kV switching, single-source scanning with split-filter, or single-source with a dual-layer detector. In contrast, photon-counting detector CT (PCD-CT) enables spectral resolution at the detector level by measuring the energy of individual photons[\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eIn neuroradiological imaging, dual-energy EID-CT minimizes image artifacts from various sources, aids in the differentiation iodine deposition from hemorrhage to guide patient management post-thrombectomy and tumor diagnosis, predicts hematoma expansion and traumatic brain injury outcomes through iodine quantification, and enhances visualization of extra-axial lesions and adjacent vessels by distinguishing hemorrhage from brain parenchymal mineralization using calcium and bone subtraction [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Furthermore, a recent review of initial clinical results of PCD-CT [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e] summarized, that PCD-CT, though in its infancy, has shown potential in reducing dosage, suppressing noise and artifacts, enhancing spatial resolution, and expanding spectral information use, which could revolutionize neuroimaging and improve clinical applications such as temporal bone imaging, stent visualization, head and neck angiographies, and stroke imaging.\u003c/p\u003e\u003cp\u003eVirtual non-contrast (VNC) images are based on a post-processing technique that uses iodine subtraction to simulate true non-contrast images of contrast-enhanced CT scans in DECT [\u003cspan additionalcitationids=\"CR7\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. However, they have demonstrated varying levels of accuracy in estimating attenuation values of the true non-contrast (TNC) images across different spectral CT technologies. Studies have shown that VNC images often closely approximate the TNC images, particularly in soft tissues, with minimal differences in attenuation values [\u003cspan additionalcitationids=\"CR10 CR11 CR12 CR13 CR14 CR15 CR16 CR17\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Nevertheless, discrepancies that emerge in tissues with higher iodine concentrations or more complicated compositions (e.g., spongy bone) can pose a challenge\u0026mdash;particularly when additional confounding materials, such as parenchymal calcifications, act as a \u0026ldquo;fourth material,\u0026rdquo; introducing beam-hardening or saturation effects that hamper VNC accuracy [\u003cspan additionalcitationids=\"CR19 CR20 CR21 CR22\" citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Larger patient size can further affect inaccuracy of VNC images, leading to greater variabilities of estimations [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. In one phantom study, VNC images for contrast materials and simulated soft tissues were found to be fairly accurate, with the exception of synthetic bone [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Despite these hurdles, improvements in photon-counting CT and dual-energy spectral CT have bolstered overall accuracy, keeping VNC errors within clinically acceptable margins [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. Nonetheless, given that the accuracy can be influenced by body habitus, different iodine concentrations, and scanner technology, clinicians should exercise caution when interpreting VNC data in routine practice.\u003c/p\u003e\u003cp\u003eThe aim of our study was to evaluate the accuracy of virtual non-contrast (VNC) images at different radiation doses and elevated iodine contrast concentrations than previously tested in a head phantom, using third generation spectral CT systems: dual-energy-integrating detector CT (TwinSpiral DECT) and photon-counting detector CT.\u003c/p\u003e"},{"header":"MATERIALS AND METHODS","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003ePhantom\u003c/h2\u003e\u003cp\u003eAn anthropomorphic acrylic CT phantom representing a human head was constructed, similar to those as in other studies [\u003cspan additionalcitationids=\"CR28\" citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. The phantom consisted of three cylindrical segments (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), each measuring 16 cm in diameter and 6 cm in height. The central cylinder contained inserts encompassing its whole height in which material simulating brain tissue and iodine solutions of different concentrations were placed. The inserts had a diameter of 1 cm and were positioned in the center and two on each side, perpendicular to each other, thus nine inserts in total. The inserts were filled with 0.9% NaCl solution, calcium carbonate (CaCO\u003csub\u003e3\u003c/sub\u003e \u0026ndash; simulating bony tissue, [\u003cspan additionalcitationids=\"CR31\" citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]), butter (simulating fatty tissue, [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]), agar (simulating brain tissue, [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]), and iodine contrast media of different concentrations (100% contrast solution (Xenetix\u0026reg; 350 mg/ml; Guerbet, Aulnay-Sous-Bois, France)), 50% contrast solution (175 mg/ml), 25% contrast solution (87,5 mg/ml) and 12.5% contrast solution (43,75 mg/ml).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eCT Imaging\u003c/h3\u003e\n\u003cp\u003eImaging was performed on two CT scanners with different underlying technologies: a single-source energy-integrating detector CT (EID-CT; Somatom X.cite, Siemens Healthcare, Erlangen, Germany) in TwinSpiral dual-energy mode [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e] and a dual-source PCD-CT equipped with two cadmium telluride PCDs (PCD-CT; Siemens NAEOTOM Alpha; Siemens Healthineers, Erlangen, Germany) [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. For each scanner, six image acquisition protocols were applied, representing varying radiation dose levels commonly used in clinical practice. From each acquisition, standard (mixed) images and virtual non-contrast (VNC) images were reconstructed for subsequent analysis.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eAcquisition and reconstruction parameters used in the TwinSpiral DECT and PCD-CT:\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParameter\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eTwinSpiral DECT (EID-CT)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003ePCD-CT\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eRotation Time\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1 s\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.5 s\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePitch\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.55\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.55\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSlice Acquisition\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e64 \u0026times; 0.6 mm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e144 \u0026times; 0.4 mm\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTube Current-Time Products (mAs)*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e220, 176, 165, 143, 110, 55\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e249, 199, 187, 162, 125, 62\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eProtocol\u0026rsquo;s CTDIvol (mGy)*\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e43.7, 35.0, 32.8, 28.4, 21.9, 10.9\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e42.0, 33.6, 31.5, 27.3, 21.0, 10.5\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMatrix Size\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e512 \u0026times; 512 pixels\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e512 \u0026times; 512 pixels\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eImaging Mode\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eDual-energy mode at 80/Sn150 kVp (tin filter for the second tube)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSpectral mode (QuantumPlus) at 120 kVp\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eReconstruction Kernel\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eQuantitative Qr40\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eQuantitative Qr40\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSlice Thickness\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e4 mm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.6 mm\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIncrement\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e4 mm\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.6 mm\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e* Tube current-time products and protocol\u0026rsquo;s CTDIvol values correspond to 100, 80, 75, 65, 50, and 25% of the standard clinical dose respectively.\u003c/p\u003e\u003cp\u003eFor each scanner, each dose setting was scanned three times, resulting in a total of 18 scans per scanner (6 settings \u0026times; 3 repetitions) and 36 scans overall (18 \u0026times; 2 scanners).\u003c/p\u003e\n\u003ch3\u003ePatients\u003c/h3\u003e\n\u003cp\u003eHead CT scans were conducted on a Somatom X.cite single-source CT scanner from Siemens Healthineers, Erlangen, Germany. The scans were performed in TwinSpiral dual-energy mode with tube voltages of 80 and 150 kVp, the latter of which utilized tin filtration. The quality reference tube current-time products were 220 and 179 mAs, respectively, and were modulated using CAREDose4D. The slice acquisition was 64 \u0026times; 0.6 mm, the rotation time was 1.0 s, and the pitch was 0.55. The protocol\u0026rsquo;s mean volume CT dose index (CTDIvol) was 43.7\u0026thinsp;\u0026plusmn;\u0026thinsp;3.4 mGy.\u003c/p\u003e\u003cp\u003eNon-contrast brain CT images from 16 patients, scanned with TwinSpiral DECT during follow-up after mechanical thrombectomy due to ischemic stroke using the above protocol, were retrospectively analyzed, HU measurements were made in the area of healthy cerebral white matter tissue, subcutaneous fat and scalp bone, and matched with standardized HU values of these tissues and phantom inserts imitating bone, brain white matter and fat. These measurements represented the values at 100% of standard dose used in clinical practice by the scanner. Only patients with no visible brain herniation or any lesions causing intracranial mass effect were included. All patients were older than 18 years. For PCD-CT no prior brain scans on patients were available. All procedures were performed in accordance with local and federal regulations and the Declaration of Helsinki. The study was approved by the local ethics committee (approval number blinded for the review purposes) and patient consent was obtained from each participant.\u003c/p\u003e\n\u003ch3\u003eQuantitative Analysis\u003c/h3\u003e\n\u003cp\u003eIn the phantom datasets, regions of interest (ROIs) were placed in the inserts containing the different materials. ROIs were drawn as large as possible (diameter\u0026thinsp;\u0026ge;\u0026thinsp;6 mm) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e), but not too close to the insert\u0026rsquo;s edges. Placement of ROIs was performed on standard (mixed) CT images and VNC images, and were then copied to the corresponding position on the other two scans. From the three ROIs of a measurement, the mean CT attenuation (HU) was recorded and an average of each three scans for each measurement was considered representative for further analyses. All measurements were repeated four times to ensure consistency. Based on the acquired parameters, an absolute and signed error vas calculated as follows:\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eVNC\u003csub\u003eerror\u003c/sub\u003e = |(HU\u003csub\u003eVNC\u003c/sub\u003e \u0026ndash; HU\u003csub\u003ebase\u003c/sub\u003e)|\u003c/p\u003e\u003cp\u003eWhere the HU\u003csub\u003eVNC\u003c/sub\u003e represents the attenuation on VNC images and HU\u003csub\u003ebase\u003c/sub\u003e corresponds to the attenuation of a standard CT (true non-contrast (TNC)) images in a phantom, respectively. () represents signed and || absolute error. Due to the occurrence of beam-hardening artifacts in some of the images, the insert with 87.5 mg/ml iodine contrast concentration was excluded from further analysis.\u003c/p\u003e\u003cp\u003eIn the patient scans, ROIs were placed in the white brain matter, cortical bone of the skull and subcutaneous fat (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Likewise, a placement of ROI was performed on respectively represented tissue inserts on standard CT and VNC images of the phantom.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003eStatistical Analysis\u003c/h2\u003e\u003cp\u003eData were first reshaped from wide to long format to capture replicate measurements. Two-way analyses of variance (ANOVAs) were then conducted on both signed errors (VNC\u0026thinsp;\u0026minus;\u0026thinsp;TNC) and absolute errors |VNC\u0026thinsp;\u0026minus;\u0026thinsp;TNC|, with radiation dose and iodine concentration as fixed factors. Whenever a main effect or interaction had a p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05, post hoc Tukey tests were performed to identify specific group differences. Residual diagnostics (QQ plots, histograms, and residuals-versus-fitted plots) were inspected to confirm that parametric assumptions (normality and homoscedasticity) were not severely violated. In addition, Shapiro\u0026ndash;Wilk and Kolmogorov\u0026ndash;Smirnov tests were used to formally test for normality of residuals. If these diagnostics suggested marked deviations from normality or large outliers, nonparametric methods (for example, Wilcoxon or other rank-based test) were considered as a backup; otherwise, the standard ANOVA approach with post hoc Tukey comparisons was retained. A two-sided p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistically significant (95% confidence interval). A Wilcoxon rank-sum was used to compare HU measurements from patients with phantom. All statistical analyses were performed in the Python programming language (version 3.14; Python Software Foundation, Wilmington, DE, USA, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.python.org/\u003c/span\u003e\u003cspan address=\"https://www.python.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e"},{"header":"RESULTS","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\u003ch2\u003ePhantom scans\u003c/h2\u003e\u003cp\u003eEffect of Radiation Dose\u003c/p\u003e\u003cp\u003eIn TwinSpiral DECT (EID-CT) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e), an ANOVA on absolute VNC errors showed a significant main effect of radiation dose (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), yet no pairwise dose comparisons were significant after Tukey adjustment (p\u0026thinsp;=\u0026thinsp;0.07\u0026ndash;1.00). Hence, none of the specific radiation dose levels (25%, 50%, 65%, 75%, 80%, 100%) differed significantly from each other in absolute errors, which ranged from 0.7\u0026thinsp;\u0026plusmn;\u0026thinsp;0.5 HU (at 65% dose) up to 2.4\u0026thinsp;\u0026plusmn;\u0026thinsp;2.2 HU (at 25% dose) when iodine concentration was 0 mg/ml, and from 8.2\u0026thinsp;\u0026plusmn;\u0026thinsp;1.1 HU (at 50% dose) up to 13.7\u0026thinsp;\u0026plusmn;\u0026thinsp;1.1 HU (at 75% dose) at 350 mg/ml. A similar ANOVA on signed errors also indicated a significant overall dose effect (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), but again no post hoc pairwise differences reached significance (p\u0026thinsp;\u0026ge;\u0026thinsp;0.05).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eIn contrast, the PCD-CT\u0026rsquo;s ANOVA (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e) on absolute errors again showed a significant main effect of dose (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), but no individual dose-level comparisons were significant (p\u0026thinsp;\u0026ge;\u0026thinsp;0.05). Across 0 mg/ml and low iodine concentrations (0 mg/ml or 43.75 mg/ml), absolute errors ranged from 1.6\u0026thinsp;\u0026plusmn;\u0026thinsp;0.9 HU (at 50% dose) up to 4.9\u0026thinsp;\u0026plusmn;\u0026thinsp;2.9 HU (at 100% dose). At 350 mg/ml, values spanned 6.8\u0026thinsp;\u0026plusmn;\u0026thinsp;1.5 HU (at 80% dose) up to 10.6\u0026thinsp;\u0026plusmn;\u0026thinsp;1.5 HU (at 25% dose). However, a separate signed-error analysis for PCD-CT revealed a specific difference between 25% and 100% radiation dose (p\u0026thinsp;=\u0026thinsp;0.0383), while other dose-level comparisons remained non-significant (p\u0026thinsp;\u0026gt;\u0026thinsp;0.05).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eEffect of Iodine Concentration\u003c/p\u003e\u003cp\u003eIn TwinSpiral DECT (EID-CT) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e), iodine concentration strongly influenced the absolute errors (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). For instance, 350 mg/ml differed significantly from 0 mg/ml (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), from 43.75 mg/ml (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and from 175 mg/ml (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Similarly, 175 mg/ml differed from both 0 mg/ml (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05) and 43.75 mg/ml (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Consequently, the absolute errors at 0 mg/ml ranged from 0.7\u0026thinsp;\u0026plusmn;\u0026thinsp;0.5 HU up to 2.4\u0026thinsp;\u0026plusmn;\u0026thinsp;2.2 HU, whereas at 350 mg/ml they reached 13.7\u0026thinsp;\u0026plusmn;\u0026thinsp;1.1 HU (75% dose). A similar pattern emerged for signed errors, which confirmed overestimations at higher concentrations: for example, +\u0026thinsp;5.0\u0026thinsp;\u0026plusmn;\u0026thinsp;0.7 HU at 175 mg/ml (100% dose) and as high as +\u0026thinsp;13.7\u0026thinsp;\u0026plusmn;\u0026thinsp;1.1 HU at 350 mg/ml (75% dose).\u003c/p\u003e\u003cp\u003eSimilarly, the PCD-CT scanner (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e) likewise exhibited a strong effect of iodine concentration, with absolute errors at 175 mg/ml and 350 mg/ml each differing significantly from 0 mg/ml (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Moreover, for signed errors, 175 mg/ml differed significantly from 43.75 mg/ml (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and 350 mg/ml differed significantly from 43.75 mg/ml (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). By contrast, the difference between 175 mg/ml and 350 mg/ml was not significant (p\u0026thinsp;=\u0026thinsp;0.19). Also, for both signed and absolute errors, 0 mg/ml vs. 43.75 mg/ml was not significant (p\u0026thinsp;=\u0026thinsp;0.058 for signed errors, p\u0026thinsp;=\u0026thinsp;0.785 for absolute errors). Across these lower iodine concentrations (0 mg/ml, 43.75 mg/ml), mean absolute errors ranged from 1.6\u0026thinsp;\u0026plusmn;\u0026thinsp;0.9 HU (50% dose) to 4.9\u0026thinsp;\u0026plusmn;\u0026thinsp;2.9 HU (100% dose). At 350 mg/ml, values spanned 6.8\u0026thinsp;\u0026plusmn;\u0026thinsp;1.5 HU (80% dose) to 10.6\u0026thinsp;\u0026plusmn;\u0026thinsp;1.5 HU (25% dose). A signed-error analysis further confirmed underestimation at higher concentrations (175 mg/ml, 350 mg/ml), which ranged from roughly \u0026minus;\u0026thinsp;2.7\u0026thinsp;\u0026plusmn;\u0026thinsp;2.5 HU (100% dose) to \u0026minus;\u0026thinsp;10.6\u0026thinsp;\u0026plusmn;\u0026thinsp;1.5 HU (25% dose).\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003ePatient scans\u003c/h3\u003e\n\u003cp\u003eA total of 16 patients (7 women, 44%, 9 men, 56%, mean age 65\u0026thinsp;\u0026plusmn;\u0026thinsp;9 years, range 51\u0026ndash;78 years) were included in the study analysis. A Wilcoxon rank-sum test showed no statistical difference (p\u0026thinsp;\u0026ge;\u0026thinsp;0.05) between average values of HU of bone or white matter between the patient and phantom images with TwinSpiral DECT (a dual-energy EID-CT), but did show significant difference of fat tissue (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eIn this study, we quantified virtual non-contrast (VNC) measurement errors in two CT scanner systems—TwinSpiral DECT (an EID-CT scanner) and photon counting detector CT (PCD-CT scanner)—across varying radiation dose levels and higher than typical iodine concentrations, simulating contrast extravasation during interventional procedures. Distinguishing blood from iodine contrast with DECT after mechanical thrombectomy allows for more precise identification of potentially complicated bleeding versus contrast extravasation. This differentiation can improve diagnostic confidence in confirming hemorrhage, better delineate the extent of blood extravasation (hematoma) separately from contrast leakage, and ultimately aid in more targeted and effective clinical management, such as surgical or interventional planning, by clearly visualizing the source and nature of the leak. We evaluated two types of error metrics: (1) absolute error, which measured how far a VNC reading deviated from a true reference value in terms of pure magnitude, and (2) signed error, which indicated whether the VNC estimate overestimated (positive sign) or underestimated (negative sign) the true value. Interpreting both metrics together was particularly valuable: absolute error captured the extent of the discrepancy, whereas signed error clarified the direction of that discrepancy.\u003c/p\u003e\u003cp\u003eOverall, the most pronounced finding was that iodine concentration strongly influenced the magnitude of VNC errors in both scanners, overshadowing many of the dose-related differences. In the TwinSpiral DECT, absolute errors were smallest (on the order of 1–2 HU) when little or no iodine was present, but rose substantially (up to 13–14 HU) as iodine concentration increased to 350 mg/ml. Likewise, in PCD-CT, errors were low (2–4 HU) at lower iodine levels and climbed to roughly 7–10 HU at higher concentrations. This pattern underscores that as more iodinated contrast remains in the image, the task of “subtracting” it to create a virtual non-contrast image becomes more challenging for both scanning systems.\u003c/p\u003e\u003cp\u003ePrevious studies similarly showed VNC accuracy being linked to iodine contrast concentration. Harsaker et al. [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e] found errors increased markedly above 50 mg/ml, escalating at 100 mg/ml. Sartoretti et al. [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e] noted a significant drop in accuracy of VNC images in EID-CT beyond 3 mg/ml, whereas in PCD-CT VNC images stayed accurate to up to 5 mg/ml. Holz et al. [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e] reported largest deviations in VNC accuracy at 5.0 mg/ml, and Mergen et al. [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e] observed stable accuracy up to 5 mg/ml. Except for Harsaker et al., however, these studies used concentrations far lower than ours, highlighting the uniqueness of our higher-concentration findings. Furthermore, it is important to note that in our study, despite the inaccuracies of VNC images at higher iodine concentrations (175 and 350 mg/ml), the visual appearance (qualitative difference) of VNC images still reflected a marked reduction in the HU compared to TNC images (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e), which may remain diagnostically useful, even if quantitative accuracy is compromised, as it might occur in vessel perforation during interventional procedure.\u003c/p\u003e\u003cp\u003eInterestingly, although the broad trend of escalating errors with higher iodine was similar in both scanners, the direction of the signed error differed. TwinSpiral DECT tended to overestimate the reference value at high iodine levels (i.e., a positive bias of roughly + 5 HU to + 14 HU), whereas PCD-CT consistently showed negative signed errors—an underestimation of the reference—ranging from around − 3 HU to − 10 HU. These opposite directions of biases likely stem from differences in each scanner’s underlying spectral decomposition algorithms, handling extreme scenarios with high iodine concentrations than usually found in soft tissue lesions, or the difference in slice thickness. Holz et al. [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e] similarly found underestimation of VNC images at 0.3 mg/ml iodine contrast concentrations, but then the directions shifted to overestimation at 5.0 mg/ml. Likewise Hua et al. [\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e] noted a shift from under- to overestimation of VNC images between 2.5 mg/ml and 5 mg/ml iodine contrast. However, these studies again used lower concentrations, underscoring the novelty of our data.\u003c/p\u003e\u003cp\u003eRegarding radiation dose, both scanners showed statistically significant main effects when analyzing absolute and signed errors across different radiation dose levels (25%, 50%, 65%, 75%, 80%, and 100% of the full dose). However, post hoc comparisons rarely produced significant pairwise distinctions. In TwinSpiral DECT, neither the absolute error nor the signed error analyses singled out any one dose level as clearly better or worse once multiple-comparison corrections were applied. Meanwhile, in PCD-CT, a single notable result emerged: at 25% vs. 100% dose, the signed error showed a statistically significant difference (p = 0.0383). Despite these findings, the overall magnitude of errors at different dose levels remained broadly comparable, and any shifts in bias were modest, although an overall slight overestimation has occurred in TwinSpiral DECT and underestimation in PCD-CT. One possible explanation is that while lower-dose scans typically exhibit increased image noise, this additional noise did not manifest as a large difference in HU measurements and thus the average VNC errors—at least not enough to become statistically distinct in most pairwise comparisons. This result suggests that moderate or even substantial dose reductions could be pursued without dramatically compromising VNC accuracy. However, the thinner slice thickness in PCD-CT compared to TwinSpiral DECT might have contributed to more noise due to greater variability of measured HU, which might have caused the significant difference between 25% and 100% radiation dose in PCD-CT. As such, careful consideration should be made when imaging with spectral CT scanner at very low radiation doses. Consistent with our findings, Li et al. [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e] reported significant inaccuracies at 75% and 50% of the clinical dose, whereas Holz et al. [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e] found no significant differences among 10, 15, and 20 mGy acquisitions. Sartoretti et al. [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e] similarly concluded that dose reductions (5, 10, 15 mGy) did not significantly affect VNC accuracy in EID-CT or PCD-CT.\u003c/p\u003e\u003cp\u003eA Wilcoxon rank-sum test showed a significant difference (p \u0026lt; 0.05) in the average HU values of fat tissue when comparing images from TwinSpiral DECT with patient data. However, due to hygienic constraints and hospital policy, we were not permitted to use real animal subcutaneous fat to represent human fat tissue. Instead, we used butter, as documented in previous studies [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e], although it does have lower HU values than actual subcutaneous fat.\u003c/p\u003e\u003cp\u003eTaken together, the current data shows that iodine concentration is the dominant factor in determining how accurately either scanner can remove or “subtract” iodine in VNC reconstructions. At lower iodine contrast concentrations, both TwinSpiral DECT and PCD-CT produced acceptably small absolute errors, but as iodine concentrations reached 175 mg/ml and above, significantly larger discrepancies from the true non-contrast baseline occurred—albeit still with a clear visual reduction in overall attenuation values. Moreover, the scanners did not err in the same direction: TwinSpiral DECT generally overshot (positive bias), while PCD-CT undershot (negative bias). Meanwhile, although the dose setting exerted a statistically significant effect in a global sense—likely reflecting differences in signal-to-noise ratios and reconstruction constraints—practical pairwise differences rarely reached significance, implying that large dose reductions could often be implemented without incurring a dramatic rise in error levels. Nonetheless, a careful balance is required: if clinicians are monitoring subtle differences in HU (e.g., a few HU changes in a small lesion), lower-dose protocols might introduce noise-related variability or small systematic biases that become clinically important.\u003c/p\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003eLimitations\u003c/h2\u003e\u003cp\u003eThere were several limitations to our study. First, for PCD-CT scanner we did not have CT head and brain images from actual patients, since at that time, the patients with neuroradiology related pathologies were not routinely scanned on PCD-CT scanner Second, the phantom used in our experiments did not incorporate manufacturer-produced tissue inserts for real-world correlation. Instead, we relied on alternative materials cited in previous research. Third, the qualitative assessment of attenuation drop on VNC images at high iodine concentrations was not conducted by a trained neuroradiologist. While the differences were so pronounced, that even a non-specialist could readily appreciate them, reducing the need for formal expert confirmation in this particular instance, we acknowledge that a formal evaluation by a neuroradiologist could have further strengthened our findings. Lastly, each scanner was used with a customized slice thickness optimized for its design. As such, instead of making a direct comparison, the results were reported separately to showcase the performance of each scanner under its specific imaging parameters.\u003c/p\u003e\u003c/div\u003e"},{"header":"CONCLUSION AND FURTHER DIRECTIVES","content":"\u003cp\u003eIn conclusion, our analysis showed that iodine concentration, rather than radiation dose, affected VNC accuracy in both TwinSpiral DECT (EID-CT) and PCD-CT, with higher concentrations leading to significant measurement deviations. Although dose reductions did not significantly degrade performance, TwinSpiral DECT tended to overestimate whereas PCD-CT tended to underestimate HU values at high iodine levels. Refining spectral decomposition algorithms and validating these findings in larger clinical cohorts could further mitigate quantitative biases and strengthen clinical utility.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eConflict of interest\u003c/p\u003e\n\u003cp\u003eSW: Nothing to declare.\u003c/p\u003e\n\u003cp\u003eRG: Nothing to declare.\u003c/p\u003e\n\u003cp\u003eZK: Nothing to declare\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eMcCollough CH, et al. Dual- and multi-energy CT: principles, technical approaches, and clinical applications. \u003cem\u003eRadiology\u003c/em\u003e. 2015;276(3):637-653. doi:10.1148/radiol.2015142631. \u003c/li\u003e\n\u003cli\u003eSo A, Nicolaou S. Spectral computed tomography: fundamental principles and recent developments. \u003cem\u003eKorean J Radiol\u003c/em\u003e. 2021;22(1):86-96. doi:10.3348/kjr.2020.0144.\u003c/li\u003e\n\u003cli\u003eLennartz S, et al. Virtual unenhanced images: qualitative and quantitative comparison between different dual-energy CT scanners in a patient and phantom study. \u003cem\u003eInvest Radiol\u003c/em\u003e. 2022;57(1):52-61. doi:10.1097/RLI.0000000000000802.\u003c/li\u003e\n\u003cli\u003eGibney B, et al. A review of the applications of dual-energy CT in acute neuroimaging. \u003cem\u003eCan Assoc Radiol J\u003c/em\u003e. 2020;71(3):253-265. doi:10.1177/0846537120904347.\u003c/li\u003e\n\u003cli\u003eAbel F, Schubert T, Winklhofer S. Advanced neuroimaging with photon-counting detector CT. \u003cem\u003eInvest Radiol\u003c/em\u003e. 2023;58(7):472-481. doi:10.1097/RLI.0000000000000984.\u003c/li\u003e\n\u003cli\u003eSauter AP, Muenzel D, Dangelmaier J, et al. Dual-layer spectral computed tomography: virtual non-contrast in comparison with true non-contrast images. \u003cem\u003eEur J Radiol\u003c/em\u003e. 2018;104:108-114. doi:10.1016/j.ejrad.2018.05.007.\u003c/li\u003e\n\u003cli\u003eAnanthakrishnan L, Rajiah P, Ahn R, et al. Spectral detector CT-derived virtual non-contrast images: comparison of attenuation values with unenhanced CT. \u003cem\u003eAbdom Radiol (NY)\u003c/em\u003e. 2017;42(3):702-709. doi:10.1007/s00261-016-1036-9.\u003c/li\u003e\n\u003cli\u003eBorges AP, Antunes C, Curvo-Semedo L. Pros and cons of dual-energy CT systems: \u0026ldquo;one does not fit all\u0026rdquo;. \u003cem\u003eTomography\u003c/em\u003e. 2023;9(1):195-216. doi:10.3390/tomography9010017.\u003c/li\u003e\n\u003cli\u003eAnanthakrishnan L, Rajiah P, Ahn R, et al. Spectral detector CT-derived virtual non-contrast images: comparison of attenuation values with unenhanced CT. \u003cem\u003eAbdom Radiol (NY)\u003c/em\u003e. 2017;42(3):702-709. doi:10.1007/s00261-016-1036-9.\u003c/li\u003e\n\u003cli\u003eJavadi S, Elsherif S, Bhosale P, et al. Quantitative attenuation accuracy of virtual non-enhanced imaging compared to true non-enhanced imaging on dual-source dual-energy CT. \u003cem\u003eAbdom Radiol (NY)\u003c/em\u003e. 2020;45(4):1100-1109. doi:10.1007/s00261-020-02415-8.\u003c/li\u003e\n\u003cli\u003eMingkwansook V, Puwametwongsa K, Watcharakorn A, Dechasasawat T. Comparative study of true and virtual non-contrast imaging generated from dual-layer spectral CT in patients with upper aerodigestive tract cancer. \u003cem\u003ePol J Radiol\u003c/em\u003e. 2022;87:e678-e687. doi:10.5114/pjr.2022.123829.\u003c/li\u003e\n\u003cli\u003eMeyer M, Nelson RC, Vernuccio F, et al. Virtual unenhanced images at dual-energy CT: influence on renal lesion characterization. \u003cem\u003eRadiology\u003c/em\u003e. 2019;291(2):381-390. doi:10.1148/radiol.2019181100.\u003c/li\u003e\n\u003cli\u003eZhang X, Zhang G, Xu L, et al. Utilisation of virtual non-contrast and virtual mono-energetic images from dual-layer spectral CT for renal cell carcinoma: image quality and radiation dose. \u003cem\u003eInsights Imaging\u003c/em\u003e. 2022;13(1):12. doi:10.1186/s13244-021-01146-8.\u003c/li\u003e\n\u003cli\u003eTakane Y, Sato K, Kageyama R, Takano H, Kayano S. Accuracy of virtual non-contrast images with different algorithms in dual-energy computed tomography. \u003cem\u003eRadiol Phys Technol\u003c/em\u003e. 2022;15(3):234-244. doi:10.1007/s12194-022-00668-0.\u003c/li\u003e\n\u003cli\u003eNiehoff JH, Woeltjen MM, Laukamp KR, Borggrefe J, Kroeger JR. Virtual non-contrast versus true non-contrast computed tomography: initial experiences with a photon-counting scanner approved for clinical use. \u003cem\u003eDiagnostics (Basel)\u003c/em\u003e. 2021;11(12):2377. doi:10.3390/diagnostics11122377.\u003c/li\u003e\n\u003cli\u003eVerstraeten S, Ansems J, van Ommen W, van Linden F, Looijmans F, Tesselaar E. Comparison of true non-contrast and virtual non-contrast images for renal lesion characterization with detector-based spectral CT. \u003cem\u003eBr J Radiol\u003c/em\u003e. 2023;96(1149):20220157. doi:10.1259/bjr.20220157.\u003c/li\u003e\n\u003cli\u003eWen D, Pu Q, Peng P, et al. Comparison of virtual and true non-contrast images from dual-layer spectral detector CT in patients with colorectal cancer. \u003cem\u003eQuant Imaging Med Surg\u003c/em\u003e. 2024;14(9):6260-6272. doi:10.21037/qims-24-535.\u003c/li\u003e\n\u003cli\u003eToepker M, Moritz T, Krauss B, et al. Virtual non-contrast in second-generation dual-energy CT: reliability of attenuation values. \u003cem\u003eEur J Radiol\u003c/em\u003e. 2012;81(3):e398-e405. doi:10.1016/j.ejrad.2011.12.011.\u003c/li\u003e\n\u003cli\u003eLehti L, S\u0026ouml;derberg M, H\u0026ouml;glund P, Nyman U, Gotts\u0026auml;ter A, Wass\u0026eacute;lius J. Reliability of virtual non-contrast CT angiography: comparing it with the real deal. \u003cem\u003eActa Radiol Open\u003c/em\u003e. 2018;7(7-8):2058460118790115. doi:10.1177/2058460118790115.\u003c/li\u003e\n\u003cli\u003eDinkel J, Khalilzadeh O, Phan CM, et al. Technical limitations of dual-energy CT in neuroradiology: 30-month institutional experience and literature review. \u003cem\u003eJ Neurointerv Surg\u003c/em\u003e. 2015;7(8):596-602. doi:10.1136/neurintsurg-2014-011241.\u003c/li\u003e\n\u003cli\u003eHarsaker V, Jensen K, Andersen HK, Martinsen AC. Quantitative benchmarking of iodine imaging for two CT spectral imaging technologies: a phantom study. \u003cem\u003eEur Radiol Exp\u003c/em\u003e. 2021;5(1):24. doi:10.1186/s41747-021-00224-2.\u003c/li\u003e\n\u003cli\u003eKanatani R, Shirasaka T, Kojima T, Kato T, Kawakubo M. Influence of beam hardening in dual-energy CT imaging: phantom study for iodine mapping, virtual monoenergetic imaging and virtual non-contrast imaging. \u003cem\u003eEur Radiol Exp\u003c/em\u003e. 2021;5(1):18. doi:10.1186/s41747-021-00217-1.\u003c/li\u003e\n\u003cli\u003eHolz JA, Alkadhi H, Laukamp KR, et al. Quantitative accuracy of virtual non-contrast images from spectral detector CT: an abdominal phantom study. \u003cem\u003eSci Rep\u003c/em\u003e. 2020;10(1):21575. doi:10.1038/s41598-020-78518-5.\u003c/li\u003e\n\u003cli\u003eLi B, Pomerleau M, Gupta A, Soto JA, Anderson SW. Accuracy of dual-energy CT virtual unenhanced and material-specific images: a phantom study. \u003cem\u003eAJR Am J Roentgenol\u003c/em\u003e. 2020;215(5):1146-1154. doi:10.2214/AJR.19.22372.\u003c/li\u003e\n\u003cli\u003eSartoretti T, Mergen V, Higashigaito K, Eberhard M, Alkadhi H, Euler A. Virtual non-contrast imaging of the liver with photon-counting detector CT: a systematic phantom and patient study. \u003cem\u003eInvest Radiol\u003c/em\u003e. 2022;57(7):488-493. doi:10.1097/RLI.0000000000000860.\u003c/li\u003e\n\u003cli\u003eMergen V, Racine D, Jungblut L, et al. Virtual non-contrast abdominal imaging with photon-counting detector CT. \u003cem\u003eRadiology\u003c/em\u003e. 2022;305(1):107-115. doi:10.1148/radiol.213260.\u003c/li\u003e\n\u003cli\u003eChou R, Li JH, Ying LK, Lin CH, Leung W. Quantitative assessment of three vendors\u0026rsquo; metal artifact reduction techniques for CT imaging using a customized phantom. \u003cem\u003eComput Assist Surg\u003c/em\u003e. 2019;24(suppl 2):34-42. doi:10.1080/24699322.2019.1649075.\u003c/li\u003e\n\u003cli\u003eFeng M, Ji X, Zhang R, Treb K, Dingle AM, Li K. An experimental method to correct low-frequency concentric artifacts in photon-counting CT. \u003cem\u003ePhys Med Biol\u003c/em\u003e. 2021;66(17):10808/1361-6560/ac1833. doi:10.1088/1361-6560/ac1833.\u003c/li\u003e\n\u003cli\u003eNomura Y, Watanabe H, Tomisato H, Kawashima S, Miura M. Gumbel distribution-based technique enables quantitative comparison between streak metal artifacts of multidetector-row CT and cone-beam CT: a phantom study. \u003cem\u003ePhys Eng Sci Med\u003c/em\u003e. 2023;46(2):801-812. doi:10.1007/s13246-023-01252-5.\u003c/li\u003e\n\u003cli\u003eBadiuk SR, Sasaki DK, Rickey DW. An anthropomorphic maxillofacial phantom using three-dimensional printing, polyurethane rubber and epoxy resin for dental imaging and dosimetry. \u003cem\u003eDentomaxillofac Radiol\u003c/em\u003e. 2022;51(1):20200323. doi:10.1259/dmfr.20200323.\u003c/li\u003e\n\u003cli\u003eCho HM, Ding H, Barber WC, Iwanczyk JS, Molloi S. Microcalcification detectability using a bench-top prototype photon-counting breast CT based on a silicon strip detector. \u003cem\u003eMed Phys\u003c/em\u003e. 2015;42(7):4401-4410. doi:10.1118/1.4922680.\u003c/li\u003e\n\u003cli\u003eCao Q, Sisniega A, Stayman JW, Yorkston J, Siewerdsen JH, Zbijewski W. Quantitative cone-beam CT of bone mineral density using model-based reconstruction. \u003cem\u003eProc SPIE Int Soc Opt Eng\u003c/em\u003e. 2019;10948:109480Y. doi:10.1117/12.2513216.\u003c/li\u003e\n\u003cli\u003eMostafavi MR, Ernst RD, Saltzman B. Accurate determination of chemical composition of urinary calculi by spiral computerized tomography. \u003cem\u003eJ Urol\u003c/em\u003e. 1998;159(3):673-675.\u003c/li\u003e\n\u003cli\u003eMenikou G, Dadakova T, Pavlina M, Bock M, Damianou C. MRI-compatible head phantom for ultrasound surgery. \u003cem\u003eUltrasonics\u003c/em\u003e. 2015;57:144-152. doi:10.1016/j.ultras.2014.11.004.\u003c/li\u003e\n\u003cli\u003eKr\u0026uuml;ger MT, Coenen VA, Egger K, Shah M, Reinacher PC. Development of a standardized cranial phantom for training and optimization of functional stereotactic operations. \u003cem\u003eStereotact Funct Neurosurg\u003c/em\u003e. 2018;96(3):190-196. doi:10.1159/000489581.\u003c/li\u003e\n\u003cli\u003eGrkovski R, Acu L, Ahmadli U, et al. Dual-energy CT in stroke imaging: value of a new acquisition technique for ischemia detection after mechanical thrombectomy. \u003cem\u003eClin Neuroradiol\u003c/em\u003e. 2023;33(3):747-754. doi:10.1007/s00062-023-01270-6.\u003c/li\u003e\n\u003cli\u003eGrkovski R, Acu L, Ahmadli U, et al. A novel dual-energy CT method for differentiating intracerebral haemorrhage from contrast extravasation after endovascular thrombectomy: feasibility and first results. \u003cem\u003eClin Neuroradiol\u003c/em\u003e. 2023;33(1):171-177. doi:10.1007/s00062-022-01198-3.\u003c/li\u003e\n\u003cli\u003eRacine D, Mergen V, Viry A, et al. Photon-counting detector CT with quantum iterative reconstruction: impact on liver lesion detection and radiation dose reduction. \u003cem\u003eInvest Radiol\u003c/em\u003e. 2023;58(4):245-252. doi:10.1097/RLI.0000000000000925.\u003c/li\u003e\n\u003cli\u003eHua CH, Shapira N, Merchant TE, Klahr P, Yagil Y. Accuracy of electron density, effective atomic number and iodine concentration determination with a dual-layer dual-energy CT system. \u003cem\u003eMed Phys\u003c/em\u003e. 2018;45(6):2486-2497. doi:10.1002/mp.12903.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Dual-energy computed tomography, photon counting computed tomography, virtual non contrast images, iodine maps, overestimation, underestimation, low-dose imaging","lastPublishedDoi":"10.21203/rs.3.rs-7345438/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7345438/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003eObjectives\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo evaluate the accuracy of virtual non-contrast (VNC) images at multiple radiation doses and high iodine concentrations using a head CT phantom with dual-energy integrating detector CT (EID-CT; TwinSpiral DECT) and photon-counting detector CT (PCD-CT).\u003c/p\u003e\u003cp\u003e\u003cb\u003eMaterials and Methods\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAn anthropomorphic head phantom containing brain tissue inserts and varying iodine concentrations (43.75, 175 and 350 mg/ml) was scanned three times with EID-CT and PCD-CT at 100%, 80%, 75%, 65%, 50%, and 25% radiation dose from which standard (mixed) and VNC images were obtained. Attenuations (HU) were measured in standard CT and corresponding VNC images, and both absolute and signed errors of VNC were calculated. For statistical analysis, data were reshaped from wide to long format. Two-way ANOVAs were conducted, considering signed and absolute errors with radiation dose and iodine concentration as factors. Post hoc Tukey tests were applied if p\u0026thinsp;\u0026lt;\u0026thinsp;0.05. Normality and homoscedasticity were checked via residual diagnostics. If assumptions were violated, nonparametric methods were used. Additionally, HU of fat, bone, and white matter from EID-CT at 100% dose were compared with 16 patient scans to confirm phantom realism.\u003c/p\u003e\u003cp\u003e\u003cb\u003eResults\u003c/b\u003e\u003c/p\u003e\u003cp\u003eANOVA showed radiation dose significantly affected VNC errors in both scanners, though only 25% vs. 100% dose in PCD-CT was significant (p\u0026thinsp;=\u0026thinsp;0.0383) after post-hoc Tukey; in EID-CT, no pairwise dose differences were significant (p\u0026thinsp;\u0026ge;\u0026thinsp;0.07). Iodine concentration exerted a stronger influence: in EID-CT, 350 mg/ml differed from 0, 43.75, and 175 mg/ml (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), producing errors up to +\u0026thinsp;13.7\u0026thinsp;\u0026plusmn;\u0026thinsp;1.1 HU. In PCD-CT, 175 mg/ml and 350 mg/ml differed from 0 mg/ml (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05) and from 43.75 mg/ml (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), showing a negative bias up to \u0026minus;\u0026thinsp;10.6\u0026thinsp;\u0026plusmn;\u0026thinsp;1.5 HU. While dose reductions had limited impact on overall VNC accuracy, higher iodine concentrations (175, 350 mg/ml) caused significant errors in both scanners, albeit with opposite signed biases.\u003c/p\u003e\u003cp\u003e\u003cb\u003eConclusions\u003c/b\u003e\u003c/p\u003e\u003cp\u003eHigh iodine concentrations caused significant VNC errors in both scanners\u0026mdash;EID-CT overestimating and PCD-CT underestimating the standard baseline\u0026mdash;yet both provided substantial visual contrast removal, with radiation dose reductions only rarely impacting accuracy.\u003c/p\u003e","manuscriptTitle":"Accuracy of virtual non-contrast images from dual-energy integrating detector CT and photon-counting detector CT at high iodine concentrations: a head phantom study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-22 08:04:10","doi":"10.21203/rs.3.rs-7345438/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"e5f79a6f-7fb2-4473-af53-fefe907ba38b","owner":[],"postedDate":"August 22nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-12-21T06:23:43+00:00","versionOfRecord":[],"versionCreatedAt":"2025-08-22 08:04:10","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7345438","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7345438","identity":"rs-7345438","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}