Feasibility of bronchial wall quantification in low‐ and ultralow‐dose third‐generation dual‐source CT: An ex vivo lung study

Abstract Purpose To investigate image quality and bronchial wall quantification in low‐ and ultralow‐dose third‐generation dual‐source computed tomography (CT). Methods A lung specimen from a formerly healthy male was scanned using third‐generation dual‐source CT at standard‐dose (51 mAs/120 kV, CTDIvol 3.41 mGy), low‐dose (1/4th and 1/10th of standard dose), and ultralow‐dose setting (1/20th). Low kV (70, 80, 90, and Sn100 kV) scanning was applied in each low/ultralow‐dose setting, combined with adaptive mAs to keep a constant dose. Images were reconstructed at advanced modeled iterative reconstruction (ADMIRE) levels 1, 3, and 5 for each scan. Bronchial wall were semi‐automatically measured from the lobar level to subsegmental level. Spearman correlation analysis was performed between bronchial wall quantification (wall thickness and wall area percentage) and protocol settings (dose, kV, and ADMIRE). ANOVA with a post hoc pairwise test was used to compare signal‐to‐noise ratio (SNR), noise and bronchial wall quantification values among standard‐ and low/ultralow‐dose settings, and among ADMIRE levels. Results Bronchial wall quantification had no correlation with dose level, kV, or ADMIRE level (|correlation coefficients| < 0.3). SNR and noise showed no statistically significant differences at different kV in the same ADMIRE level (1, 3, or 5) and in the same dose group (P > 0.05). Generally, there were no significant differences in bronchial wall quantification among the standard‐ and low/ultralow‐dose settings, and among different ADMIRE levels (P > 0.05). Conclusion The combined use of low/ultralow‐dose scanning and ADMIRE does not influence bronchial wall quantification compared to standard‐dose CT. This specimen study suggests the potential that an ultralow‐dose scan can be used for bronchial wall quantification.


| INTRODUCTION
Airway quantification on computed tomography (CT) provides an estimate of bronchial remodeling and inflammation, which is associated with physiological parameters and symptoms in chronic obstructive pulmonary disease (COPD). [1][2][3][4][5][6] The increasing concern on radiation exposure in clinical practice stimulates the new techniques to decrease radiation dose, 7 like peak kilovolt (kV) reduction and iterative reconstruction (IR). 8 Iterative reconstruction has been widely used for reducing radiation dose while improving image quality. [9][10][11][12] Recently, due to the technical advances and natural tissue contrast between lung parenchyma and airway, the radiation dose of chest CT is possible to closer to that of chest x-ray (approximately < 0.1 mSv). 13,14 Whether the lower radiation dose affects the evaluation of COPD compared with conventional-dose CT has become an important issue in COPD studies. Some studies showed no significant side effect of dose reduction on emphysema evaluation. [15][16][17] For example, ultralow-dose CT at a radiation dose equivalent to 5% of the standard dose (2.33 ± 1.54 mSv), although increasing the percentage of low attenuation area, was strongly correlated with standard-dose CT, thus could be used to evaluate lung volume and density. 15 However, the reduction in radiation dose decreases the ability to display distal bronchi. Kirby et al. showed that the spatial resolution of 1-2 mm in low-dose CT (about 1.5 mSv) could identify the bronchi with an inner diameter greater than 2.5 mm. 18 While CT examination at higher radiation dose (11.2 mSv) could show small airways with a diameter of 0.8 mm. 19 In an ex vivo porcine lung specimen study when radiation dose decreased, the number of evaluable bronchial branches decreased, but measurement variability increased. 20 To the best of our knowledge, there are few studies investigating bronchial wall quantification using low/ultralow-dose CT in the human lung.
Low kV settings and advanced modeled iterative reconstruction (ADMIRE) have been recently introduced, available on third-generation dual-source CT systems. 21 Therefore, this study aimed to comprehensively validate the correlation between bronchial wall quantification and low/ultralow-dose CT techniques, using an ex vivo human lung specimen.

2.A | Specimen
An ex vivo lung from a formerly healthy nonsmoking male was used to evaluate the accuracy of bronchial measurements. The specimen size was about 19.5 cm long, 9.5 cm wide, and 7.5 mm high. The anatomy department of the medical school provided the specimen for research purposes.
The Sn setting involves the prefiltration of the X-ray beam by using a tin filter. This filter limits the range of the X-ray energy spectrum reaching the scanned object. 25 The tube current was adjusted to fit the predefined CTDI vol for each tube voltage (Table 1). Other settings were kept constant across standard-and low/ultralow-dose scans: matrix size 512 × 512, CARE Dose4D on, detector collimation 192 × 0.6 mm, rotation time 0.5 s (a slower gantry rotation speed for higher image quality), 26 slice thickness 0.6 mm, and field of view 177 mm (cover all the specimen). All the images were reconstructed using Qr40 kernel (a kernel especially for quantitative analysis) with ADMIRE levels 1, 3, and 5, including the full-dose scan. We repeated each scan five times and performed a small translocation and rotation of the specimen in between to simulate variability among repeated clinical scans. Due to certain differences between the specimen and human lung, a few of the scans could not be reconstructed by software, therefore, airway data of the corresponding repeated scan were excluded from the analysis, while SNR and noise data were retained.    should be close to a round shape. B1 was in the left lower lobe bronchi, B2 in the next generation one third of the proximal bronchi, and so on. We marked the location of B1-B7 in Fig. 1(b). The software automatically measured average wall thickness (WT) and wall area percentage (%WA). %WA was defined as (wall area)/(wall area + lumen area) × 100%. We showed multiple representative cross-sectional images explaining how to segment and measure the bronchial wall in Fig. 2.

2.C | Evaluation method
The measurement points across different scans were the same distance between the point and the left main bronchus bifurcation.
Background signal-to-noise ratio (SNR) and noise were measured by placing a circular region of interest (ROI; area of 20 mm 2 ) in the lumen of the left main bronchus.

2.D | Statistical analysis
The relationship between measurement values (SNR, noise, WT, and %WA) and protocol settings was evaluated by Spearman correlation analysis. The protocol settings consisted of radiation dose levels ADMIRE levels were pairwisely evaluated using the Student-New-

3.A | Overview
In total, 108 sets of image reconstruction were accomplished. All the images did not show blocky pixelated at any ADMIRE level (Fig. 3).
The semi-automated software accurately segmented the external and

3.D | Pairwise comparisons of bronchial wall quantification in different dose settings
There were no significant differences in WT and %WA between each of the low/ultralow-dose settings and standard-dose setting or in pairwise comparison for the different low/ultralow-dose settings (all P > 0.05; Table 3).
T A B L E 2 Pairwise comparisons of signal-noise-ratio (SNR) and noise for low/ultralow-dose settings versus standard-dose setting

| DISCUSSION
Low-dose chest CT has been clinically implemented for more than 20 yr. 27 Recently introduced ultralow-dose CT with a radiation dose similar to chest X-ray (approximately 0.15 mGy or 0.06 mSv) has been applied to pulmonary diseases. 14,17,28 Studies for low/ultralowdose CT mainly focused on validation of image quality, 13,23,24,29 disease detection, and dose reduction. 25,30 Its application in thoracic imaging mainly on diagnostic confidence and detectability of pulmonary nodules, 13,14,22,28,31,32 only a few on interstitial pulmonary disease, 24 pulmonary inflammation, 14 emphysema evaluation 17 , and airway assessment. 20,33 The growing morbidity and mortality of COPD has given rise to suggestions for screening using low/ultralow-dose CT in the at-risk population for preventive treatment, 34 and to identify specific subgroups and exacerbation which may be amenable to therapy. 1 Therefore, we studied the image quality using low/ultralow-dose CT, low kV settings, and ADMIRE levels, and analyzed the impact on bronchial wall quantification. We found that SNR and noise were significantly influenced by dose levels and ADMIRE levels, but not significantly influenced by kV settings. Dose, kV settings, and ADMIRE levels showed no obvious influence on bronchial wall quantification.
Radiation dose, image quality, and diagnostic accuracy in low/ultralow-dose CT need to be balanced. There is a disagreement on whether low/ultralow-dose scanning affects CT measurements such as emphysema index, 16,17 related to reduced image quality. A recent study showed that advanced CT techniques like tin filtration and IR were able to generate CT images with acceptable noise for quantification at ultralow-dose (CTDI vol of 0.15 mGy) in COPD patients. 13 For airway measurements, an ex vivo study showed that a low radiation dose (minimum of 0.25 mGy) did not influence measured airway parameters using IR. 20 An in vivo large animal study showed that ultralow-dose CT protocols had small measurement differences of % ADMIRE allows to decrease radiation exposure by retrospectively eliminating increased image noise, and therefore retaining image quality. 23,30 The ADMIRE level mainly controls the strength of noise reduction. Therefore, an increasing ADMIRE level should allow higher radiation dose reduction. 9 In our study, ADMIRE 5 resulted in improved SNR and lower noise in comparison to ADMIRE 1 and 3 with the same radiation dose, which confirms results from previous studies. 13,36 Although the ADMIRE strength of 5 possibly had a higher amount of noise reduction, blocky appearance (losing detail in the image) of higher strength IR 30,37 was not observed in our study, which may due to ADMIRE's excellent noise reduction potential, the technology adopts a "statistical model" and iterative decoding chip to integrate the statistical data of virtual data domain, image domain, and model domain efficiently. 38 Through multiple iterations, the artifact was removed and the noise was reduced to achieve real-time high-definition iterative imaging. 38 25 In another anthropomorphic chest phantom study, Sn100 kV was found to be better than 70 kV for nodule detection and noise reduction in low/ ultralow-dose CT using ADMIRE 3 and 5. 28 A study of 1/10 dose (0.32 mSv) using Sn100 kV showed that subjective image quality was not statistically significantly different from the standard 3 mSv dose group. 29 In our study, image SNR and noise values and airway quantification values showed no obvious changes among kV settings at the same dose and the same ADMIRE level, although Fig. 4 showed a slight noise reduction in the 1/20 ultralow-dose using Sn100kV and ADMIRE 5. In terms of diagnosis, low kV showed no significant difference from the standard dose in the bronchial wall measurement. We speculated that the result may be related to the scanned object.
This study has limitations. First, although the ex vivo lung lacks radiation absorption by the thoracic cage, similar human lung specimens for structural evaluation have proved its usability for research. 34 We used CARE Dose4D to adjust tube current, and thus, exposure dose. Second, the measurements using standard dose were considered as the reference standard because it was impossible to dissect the lung specimen to obtain true values of the bronchial.
Third, CTDI vol is an indirect measure but a surrogate indicating the radiation output of the CT system. 21 Fourth, our results need to be confirmed in clinical patient studies with regard to the impact on quantitative emphysema assessment before implementation in COPD patients.

| CONCLUSION
Ultralow-dose settings increased image noise and slightly increased measurement variability, combined higher ADMIRE compensated for the increased noise caused by low-dose while did not significantly influence the bronchial measurements. This specimen study suggests that an ultralow-dose scan as low as 0.17 mGy is useful for bronchial wall quantification. If a patient study confirms these findings, ultralow-dose settings could be used in COPD patients.
T A B L E 4 Agreement of wall thickness (WT) and wall area percentage (%WA) between each of low/ultralow-dose settings and standard-dose setting evaluated by Bland-Altman analysis WT represents wall thickness; %WA represents wall area percentage; SD represents standard deviation; and CI represents confidence interval.