The study was conducted in accordance with the European Directive 2010/63/EU and with the German law for animal protection (TierSchG). All experiments were approved by the local animal ethic committee (Lower Saxony State Office for Consumer Protection and Food Safety, LAVES 18/2809). Seven female domestic pigs weighing 68−88 kg were used. Anaesthesia was induced by intramuscular injection of 10 mg/kg tiletamine/zolazepam (Zoletil®, Virbac, Switzerland) and atropine (Atropinsulfat, BBraun, Germany) followed by intravenous injection of 10 mg/kg of propofol (Narcofol®, CP Pharma, Germany) to enable endotracheal intubation. Animals were maintained under general anaesthesia using an isoflurane precision vaporiser and mechanically ventilated (air-oxygen mixture 1:1; pIso > 1.8 mmHg). The breathing rate was set to 12 breaths per minute and a ventilation volume of 8−10 mL/kg per breath was chosen, based on the end-tidal CO2 (35−45 mmHg) concentration. The depth of anaesthesia was continuously monitored (ECG, capnography, body temperature, blood pressure, and O2 saturation). Animals received continuous fluid therapy (Ringer’s lactate, 10 nmg/kg/h), had a urinary catheter, and were positioned on a warming pad to maintain body temperature. Analgesia was achieved by an initial systemic intravenous dose of 4 mg/kg of carprofen (Rimadyl®, Zoetis, USA). Lidocaine (1 mg/kg, Xylocain, Aspen, Germany) was locally infiltrated at the arterial access sites. To obtain series of sharp images without movement of the diaphragm, the animals received an intravenous injection of 0.1 mg/kg of pancuronium (Pancuronium 2 mg/mL, Rotexmedica GmbH, Germany) every 1.5 to 2 h during anaesthesia. At the end of the experiment, the pigs were euthanised under deep anaesthesia by intravenous injection of 15–25 mL/animal of T61 (MSD, Unterschleißheim, Germany) until heart arrest was confirmed.
After local anaesthesia, a 5F introducer sheath was inserted into both femoral arteries under sonographic guidance. Appropriate angiographic catheters were then used to catherise the liver and the kidney. Embolisation of an appropriate subsegmental liver and kidney artery was conducted using tantalum-based embolisation material (Onyx®, Medtronic, Tolochenaz, France). In order to simulate iodine injections, an adjacent subsegmental liver and kidney artery occlusion was conducted with Lipiodol (Guerbet, Sulzbach, Germany) and cyanoacrylate glue (Histoacryl, B|Braun, Rubi, Spain) using a 4:1 ratio. A 3F Thru-Lumen embolectomy Fogarty catheter (Edwards Lifesciences Corp, Irvine, USA) filled with Iomeprol 300 (Bracco Imaging, Konstanz, Germany) was placed in the aorta holding a 0.0018” steel guidewire (Ashaee 18 guidewire, ASAHI INTECC CO LTD, Aichi Japan; V18TM Control Wire, Boston Scientific, Ratingen, Deutschland). Furthermore, a 0.0014” micro guide wire (Tenor, Transcend, MeritMedical, Jordan UT, USA) within a microcatheter (Maestro, MeritMedical, Jordan, UT, USA) and 5F guide catheter was placed in a liver artery. Overall, nine imaging scenarios were created in the seven animals (Fig. 1).
In order to simulate different patient absorptions, doublets of polymethyl-methacrylate plates (30 × 30 × 2 cm3), combined with aluminium plates (30 × 30 × 0.2 cm3) were used to simulate 25 mm of soft and bone equivalent tissue. Two and four doublets were placed below the examination table using a custom-made holder and thus simulated additional 5 cm or 10 cm soft and bone equivalent tissue .
Imaging was conducted using a robotic C-arm angiography system (ARTIS pheno®, Siemens Healthcare GmbH, Forchheim, Germany), incorporating both types of exposure control (CEC and DEC). Fluoroscopic images of each scenario were acquired in one session using three clinically established DEC protocols (Table 1) with three different dose levels—low dose (LD); normal dose (ND); high dose (HD)—followed by the corresponding Ta-specific CEC protocols (Table 2). The acquisition of one scenario took about 4 to 6 min. Considering the different operating principle of CEC as compared to DEC, CEC was parametrised by different settings. CEC aims for a predefined material-specific, spatially and temporally frequency-dependent CNR instead of a constant detector dose at the image receptor. Image quality of CEC was parametrised by the following three parameters: the IQ level, which is proportional to CNR2; the IQ gradient; and the reference water value, at which the specified IQ level should be reached if IQ gradient is not 0. The IQ gradient allows adjusting the IQ dependent on the water equivalent thickness (WET), which is constant for an IQ gradient = 0 and increases strongest towards thinner regions for an IQ gradient = 1.0. In this study an IQ gradient = 0.5 was used, as this provides a nearly constant image noise . The IQ levels were set in small increments to provide a close comparison of IQ and Ka,r of fluoroscopic images acquired using CEC with the fluoroscopic images acquired using DEC (Fig. 2). Scenes were acquired with the animal placed in the isocentre, a field of view of 42 cm, and a source to image distance of 110 cm. The acquisition time of one fluoroscopic scene covered approximately 75% of a breathing cycle. Due to the different acquisition characteristics of both exposure controls, the image acquisition protocols were first optimised to determine the appropriate image acquisition parameters. Finally, six imaging scenarios acquired with DEC and CEC were compared in this study.
Postprocessing was minimised in order to reduce its possible effects and contained only a quadratic base curve and a gamma correction. All other post processing such as edge enhancement, noise reduction, or temporal averaging was switched off by using the service mode. The images were transferred for further evaluation to the Picture Archiving and Communication System. Images contained all of the image acquisition parameters, the WET of the projection, and the Ka,r measured using the dose-area product meter, which is incorporated into the x-ray source and the collimator assembly [11, 12].
Image quality assessment
Dose equivalent imaging
First, the Ta-CEC acquired (dose equivalent Ta image) fluoroscopic image with an equivalent Ka,r compared to the DEC acquired fluoroscopic image (DEC-image) was selected. The dose equivalent Ta image was defined as the image with a Ka,r closest to the Ka,r of the DEC image and whose Ka,r was no greater than 5% above the Ka,r of the DEC image. The image dataset containing the LD, ND, or HD DEC image with the respective dose equivalent Ta image were presented on a workstation (Visage 7.1.15, Visage Imaging, Berlin, Germany) to three blinded readers with 5, 8, and 9 years of experience in angiography (Fig. 1). A digital radiograph of the respective Ta structure served as reference. The readers evaluated the IQ of the Ta structure using the five point Likert scale given in Table 3.
Image quality equivalent imaging
In a second step, the fluoroscopic images acquired with Ta-CEC and different IQ levels were compared by the readers to the fluoroscopic images acquired with DEC and different dose levels. The readers were asked to select the fluoroscopic image acquired with Ta-CEC that most closely provided the IQ acquired with DEC (Ta-IQ equivalent image). For each image dataset, the radiologist specifically focused on the depiction of the tantalum structures. The Ka,r of the Ta-IQ equivalent image was subsequently documented and compared to the Ka,r of the corresponding DEC image (Fig. 2).
Descriptive statistical analyses were calculated (mean value ± standard deviation) [13,14,15]. Interobserver agreement between the three readers was calculated using the two-way random intraclass correlation coefficient with absolute agreement. The following classification was used for interpreting the agreement: poor (< 0.40); fair (0.40−0.59); good (0.60−0.74); excellent (≥ 0.75) . The comparison of IQ ratings of the DEC image with IQ rating of the corresponding dose equivalent Ta image was performed using the visual grading characteristics (VGC) analyser software . In VGC analysis, IQ ratings for two different conditions are compared by producing a VGC curve, similar to how the ratings for normal and abnormal cases in receiver operating characteristic analysis are used to create its curve. The software computes the area under the visual grading curve (AUCVGC) with 95% confidence interval using bootstrapping (n = 2,000), taking possible dependencies between ratings into account. An AUCVGC value of 0.5 reflected similar IQ for the two exposure controls, while an AUCVGC > 0.5 indicated superior IQ, and an AUCVGC < 0.5 indicated inferior IQ produced by CEC.
Possible Ka,r differences between the DEC-image and the dose equivalent Ta image were assessed via paired t test according to Norman  who demonstrated the robustness of parametric tests. To assess the dose differences between the DEC image and the respective Ta-IQ equivalent image, the mean of the Ka,r values selected by the three readers was first calculated and then compared with the paired t test. Normal distribution of the dose data was confirmed level-wise using the Shapiro-Wilk test. Differences in Ka,r reduction between the different dose levels were assessed with the ANOVA test and post-hoc paired t test with Bonferroni correction. A p value < 0.05 indicated significance.
The required sample size was conservatively estimated, based on the estimated dose difference of at least 30 ± 20% with an abnormal distribution of the paired acquisitions. This requires at least six scenarios to achieve a power of 0.8 at an α error of 0.05 . The statistical analyses in addition to the VGC software were performed with R (R version 3.6.3, http://www.r-project.org with package “IRR” version 0.84.1) and the sample size estimation with G*Power (G*Power18.104.22.168, Kiel, Germany).
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.