This research was prospectively performed in a single medical center. Written informed consent was obtained from all the patients. The prospective study protocol was approved by the Institutional Ethics Committee (No. HS-2427) On June 23, 2020.

Sample size estimates were calculated using PASS version 15.0 (Kaysville, Utah, USA), where the module of “one-way repeated measures” was carried out on a pilot study of 10 patients. The prestudy indicated that only a dozen patients were needed to satisfy the power of 90% with a Wilks Lambda Test at a 0.05 significance level.

From June 2020 to October 2022, consecutive adult patients who were clinically diagnosed with LVV (either TAK or GCA) and referred for imaging assessment of the aorta and its branches were enrolled in our study. The exclusion criteria were those aged < 18 years, with contrast-related allergy, impaired renal function, pregnancy, thyrotoxicosis, and inadequate imaging quality. Seventy-two consecutive adult patients were initially enrolled in our study, and 50 were finally included. The mean age of the included 50 patients (43 women, 86%) was 33.5 ± 11.4 (range 18–60) years.

All aortic CTA were performed using a 320-row-detector CT scanner (Aquilion ONE GENESIS Edition; Canon Medical Systems Corp., Otawara, Japan). The scanning sequences included non-contrast, arterial, and delayed phases, and the parameters were as follows: tube voltage, 100 kVp; rotation time, 0.5 s; tube current adjusted automatically with a noise index of 7.5; collimation, 100× 0.5 mm; field of view, 400 mm. All patients received 50–60 mL of IV iodinated contrast agent (iopamidol injection, 370 mg I/mL, Shanghai Bracco Sine Pharmaceutical Corp.) at a rate of 4 mL/s via the antecubital vein using a dual-syringe power injector (Nemoto-Kyorindo, Tokyo), followed by 30–40 mL of saline at the same injection rate. The images were acquired from 2 cm above the thoracic entrance to 2 cm below the lower margin of the pubic bone.

A bolus tracking technique was applied with a trigger threshold of 180 HU in the descending aorta. Arterial phase imaging was initiated. Delayed-phase imaging was performed for 70 s after contrast medium injection.

Arterial- and delayed-phase images were reconstructed using Hybrid-IR (Adaptive Iterative Dose Reduction 3D, FC08 kernel, Canon Medical Systems; hereafter, hybrid iterative reconstruction (HIR)) and DLR (AiCE, Body Sharp Kernel, Canon Medical Systems), with a slice thickness of 1.0 mm and an interval of 0.8 mm. The corresponding DB CTA images were generated using CE-Boost software (SURESubtraction, Canon Medical Systems) to obtain the DB-HIR and DB-DLR image datasets.

Two radiologists (with 9 and 19 years of experience in reviewing CTA studies) independently performed qualitative image analysis using a dedicated workstation (Advantage Workstation 4.7; GE Healthcare). One dedicated slice with maximum wall thickening for each patient was selected from the ascending aorta, aortic arch, descending thoracic aorta, or abdominal aorta. The two radiologists independently reviewed the images and were blinded to the type of image reconstruction algorithm used. The six image datasets evaluated included the arterial phase (A-HIR and A-DLR), delayed phase (D-HIR and D-DLR), and DB (DB-HIR and DB-DLR). The initial window width and level were set to 350 and 50 HU, respectively; both parameters were modifiable. Three subjective parameters were evaluated: –outer-wall delineation, inner-wall delineation, and overall image quality. A 4-point scale evaluated outer-wall delineation and inner-wall delineation: 1 = not identified, < 25% circumscribed margins indistinct density between the outer wall and peri-aortic fat (inner wall and lumen); 2 = poorly identified, 25%–50% circumscribed margins somewhat indistinct; 3 = fairly identified, 50–75% circumscribed margins somewhat distinct; 4 = being well identified, > 75% circumscribed margins distinct. The overall image quality was assessed by scoring the interpretability of the aortic wall using a four-point Likert scale: 1 = non-diagnostic, 2 = moderate visualization, 3 = good visualization, and 4 = excellent visualization [22].

Aortic wall thickness was measured on the same axial slice and workstation used for the qualitative evaluation of image quality. Two reviewers independently measured the maximum aortic wall thickness using A-HIR, A-DLR, D-HIR, D-DLR, DB-HIR, and DB-DLR.

All CT images were manually segmented using the open-source software ITK-SNAP (http://www.itksnap.org/pmwiki/pmwiki.php) to obtain the boundaries of the lumen and inner and outer walls of the aorta. Four regions of interest were placed in the slice with maximum wall thickening and carefully outlined by one radiologist with 9 years of experience, encompassing different structures: the lumen, inner wall of the aorta, outer wall of the aorta, and peri-aortic fat. Quantitative parameters, including the mean, standard deviation (SD), signal-to-noise ratio (SNR), and contrast-to-noise noise ratio (CNR), were automatically calculated using Python code (version 3.6.4). The SNR of the vessel wall, CNR between the wall and lumen (CNRinner), and CNR between the wall and peri-aortic fat (CNRouter) were calculated as follows:

$$}}}}=\frac}}}}}_}}}}}}}}}}}_}}}}}}$$

$$}}}}}_}}}}}=\frac}}}}}}}_}}}}}-}}}}}_}}}}}}}}(}}}}}_}}}}}^+}}}}}_}}}}}^)}}$$

$$}}}}}_}}}}}=\frac}}}}}}}_}}}}}-}}}}}_}}}}}}}}(}}}}}_}}}}}^+}}}}}_}}}}}^)}}$$

The CT dose index (CTDIvol) and dose-length product (DLP) were obtained for each patient. The effective radiation dose (ED) was calculated as the DLP multiplied by a conversion factor of 0.015 (mSv · mGy−1 · cm−1) [23].

Statistical analyses were performed using R software (version 3.6.1; http://www.R-project.org). Quantitative data are expressed as the mean ± SD or median and interquartile range, when appropriate. The Shapiro–Wilk test was used to assess the normal distribution of the data. For continuous variables with a normal distribution, one-way ANOVA was used, and pairwise t-tests with Bonferroni correction were used for multiple comparisons. The Friedman test was applied to non-normally distributed data, and the Wilcoxon signed-rank test with Bonferroni correction was used for multiple comparisons. Pairwise comparisons of HIR and DLR were used to illustrate the impact of the reconstruction algorithms, whereas pairwise comparisons between DB and other phases were performed to demonstrate the value of the DB technique. Statistical significance was set at p < 0.05.

Inter-observer agreement of qualitative image analyses was assessed by kappa coefficients, and that of vessel wall thickness measurement was calculated by the intraclass correlation coefficient (ICC) (criteria: ≤ 0.40, poor; 0.41–0.60, moderate; 0.61–0.80, good; > 0.80, excellent).