Study Population
Patients were eligible for inclusion if they were 0–18 years of age and had undergone a cardiac catheterization procedure with 3DRA acquisition at the Wilhelmina Children’s Hospital between October 2014 and October 2015. The institutional review board approved this study and no informed consent was required. Retrospective analysis of medical records and catheterization data was performed. Parameters collected include age, weight, height, body surface area (BSA), cardiac diagnosis, and type of intervention (if applicable). Patients were grouped according to their initial diagnosis. Patient characteristics of this low-dose cohort 3DRA were compared with a patient group previously reported, undergoing a normal-dose 3DRA [14]. Reasons for exclusion from ED calculation were incomplete rotation, wrong positioning of the patient, and insufficient contrast. As contrast absorbs radiation, insufficient contrast leads to less radiation exposure. 3DRAs made with a central venous catheter or because of a non-cardiac diagnosis were excluded from image quality assessment.
3DRA Image Acquisition
3DRAs were obtained using the Siemens Artis Zee biplane (Siemens, Forchheim, Germany) and reconstructions were transferred to the Leonardo workstation for post-processing with Syngo DynaCT Cardiac software. All procedures were performed under general anesthesia. Rapid atrial or ventricular pacing was performed in 88 of the 100 3DRAs. Pacing frequency was increased from 180/min upwards until a reduction of 50% of the systolic blood pressure was achieved. Contrast medium was administrated to the cardiac compartment prior to the region of interest meaning the right ventricle for pulmonary imaging and the left ventricle for aortic imaging. Contrast was diluted up to 60% with saline. Contrast was injected from 2 mL/s in 3 kg neonates up to 16 mL/s in 50 kg adolescents in case of a single injection site before start of 3DRA for 5 s. When multiple injection sites were necessary, additional manual injections with diluted contrast were performed.
Dose Reduction
Compared to the study of Peters et al., the number of frames per second was reduced from 60 to 30 f/s [6]. In addition, patients were scanned with a tube voltage corresponding to a low-dose program (Table S1). All patients were scanned according to a protocol of one weight class lower than the patient’s weight. Furthermore, a thick copper filter was used to filter out low-energy photons that can cause harm and do not contribute to the image quality. Collimation was applied with a diaphragm to protect irrelevant tissue from radiation and to prevent scattering rays from causing background haze. Before the actual run, tube current was checked to be below 100 mA. If not, image plane was checked for metal artifacts and the tube current was automatically adjusted accordingly.
Calculation of ED
All data required for calculation were extracted from Artis Zee biplane and converted to Microsoft Excel (Microsoft, USA) with CareAnalytics (Siemens, Erlangen, Germany). Parameters describing the geometry of the X-ray tube, the radiation quality (tube current, filter material and thickness, and anode angle), and the patient (age, height, and weight) were imported in Monte Carlo program PCXMC 2.0 (STUK, Finland) from Microsoft Excel to calculate ED. The outcomes of the Monte Carlo stimulations are according to International Commission on Radiological Protection 103 (ICRP103) organ weighing factors [15, 16].
Image Quality Assessment
3DRA images were extracted from the Leonardo workstation and patient characteristics and cohort information were removed from the files. One experienced pediatric cardiologist and two junior researchers blindly assessed image quality independently. Pre-defined anatomical structures, necessary for diagnosis or intervention, were separately scored on a three-point scale (good = 3, moderate = 2, and poor = 1).
Statistical Analysis
Continuous values were expressed as median with a range, and categorical values as a number and percentage of the total. Differences between baseline characteristics were tested for significance using T-test or 2-tailed Mann Whitney test, for normally distributed and skewed continuous values, respectively. Significant differences for gender, diagnosis, and image quality were tested with a Chi-square test. P < 0.05 was considered to be significant. A Kruskal–Wallis test was used to compare age, weight, and ED per initial diagnosis. Spearman’s correlation testing and linear regression were performed to evaluate the association of ED with patient’s age, height, weight, BSA, skin dose, DAP, and tube current in case of non-normally distributed variables. A Fleiss kappa was calculated to test interobserver agreement of the image quality. All statistical calculations were performed using Microsoft Excel 14.6.1.