Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Emphysema quantification using chest CT: influence of radiation dose reduction and reconstruction technique

Emphysema quantification using chest CT: influence of radiation dose reduction and reconstruction... Background: Computed tomography (CT) emphysema quantification is affected by both radiation dose (i.e. image noise) and reconstruction technique. At reduced dose, filtered back projection (FBP) results in an overestimation of the amount of emphysema due to higher noise levels, while the use of iterative reconstruction (IR) can result in an underestimation due to reduced noise. The objective of this study was to determine the influence of dose reduction and hybrid IR (HIR) or model-based IR (MIR) on CT emphysema quantification. Methods: Twenty-two patients underwent inspiratory chest CT scan at routine radiation dose and at 45%, 60% and 75% reduced radiation dose. Acquisitions were reconstructed with FBP, HIR and MIR. Emphysema was quantified using the 15th percentile of the attenuation curve and the percentage of voxels below -950 HU. To determine whether the use of a different percentile or HU threshold is more accurate at reduced dose levels and with IR, additional measurements were performed using different percentiles and HU thresholds to determine the optimal combination. Results: Dose reduction resulted in a significant overestimation of emphysema, while HIR and MIR resulted in an underestimation. Lower HU thresholds with FBP at reduced dose and higher HU thresholds with HIR and MIR resulted in emphysema percentages comparable to the reference. The 15th percentile quantification method showed similar results as the HU threshold method. Conclusions: This within-patients study showed that CT emphysema quantification is significantly affected by dose reduction and IR. This can potentially be solved by adapting commonly used thresholds. Keywords: Densitometry, Emphysema, Radiation dosage, Thorax, Tomography (x-ray computed) Key points was achieved at 75% reduced dose with hybrid iterative reconstruction Dose reduction resulted in a significant CT overestimation of emphysema, while iterative Background reconstruction resulted in a significant Chest computed tomography (CT) offers the possibility of underestimation quantifying the amount of emphysema. The number of This can potentially be solved by adapting the chest CT acquisitions is expected to increase due to the commonly used densitometry thresholds favourable results of the National Lung Screening Trial [1] The maximal intraclass correlation coefficient and the interest in subtyping chronic obstructive pulmon- between reduced dose and the reference standard ary disease (COPD) patients [2]. Additional quantification of emphysema on screening CT acquisitions will therefore likely gain importance. Furthermore, this additional infor- * Correspondence: a.m.denharder@umcutrecht.nl mation may also contribute to optimisation of the benefits Department of Radiology, University Medical Center Utrecht, Utrecht, The and cost-effectiveness of CT screening [3]. CT can be used Netherlands Full list of author information is available at the end of the article to both identify patients with emphysema as well as to © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. den Harder et al. European Radiology Experimental (2018) 2:30 Page 2 of 10 monitor progression in patients with COPD. Although em- vivo information about pathological changes and allows physema is traditionally a pathology-based diagnosis [4, 5], for differentiation between airway obstruction and em- CT densitometry of the lungs has demonstrated it to be as- physematous destruction [10]. sociated with airflow obstruction, forced expiratory volume The increasing number of chest CT scans has urged in 1 s and severity according to the Global initiative for the importance of radiation dose reduction. However, chronic Obstructive Lung Disease (GOLD) criteria [6–8]. dose reduction leads to higher noise levels, especially CT densitometry is based on either the 15th percentile of when images are reconstructed using conventional fil- the attenuation curve or the percentage of voxels below tered back projection (FBP). Therefore, several iterative -950 HU because those parameters show the strongest reconstruction (IR) techniques were developed to reduce correlation with microscopic and macroscopic emphy- image noise [11, 12]. Recent studies showed that the ra- sema findings [8, 9]. Although pulmonary function tests diation dose of unenhanced chest CT can be reduced to measure limitation of airflow, they are not able to dif- sub-millisievert dose levels when IR is applied [13]. ferentiate between airway obstruction and emphysema- It is known that CT emphysema quantification is af- tous destruction. CT, on the other hand, provides in fected by both radiation dose (i.e. image noise) and Fig. 1 Example of the semi-automatic software which was used for emphysema quantification. First, the airways, lungs and lung lobes are segmented (a). Subsequently, a histogram is made which displays the number of voxels with a certain density (b). In this example the percentage of voxels below -950 HU is displayed den Harder et al. European Radiology Experimental (2018) 2:30 Page 3 of 10 reconstruction technique. At reduced dose, FBP results weight ≥ 80 kg). The tube current-time product was 60 in an overestimation of the amount of emphysema due mAs at routine dose and subsequently decreased to 33, 24 to higher noise levels, while the use of IR can result in and 15 mAs to achieve 45%, 60% and 75% dose reduction, an underestimation due to reduced noise [14–16]. respectively. All four acquisitions were acquired consecu- The primary aim of the current study was to determine tively in a single session. Automatic exposure control was the effect of both dose reduction and IR on CT emphy- off. Images were reconstructed at a slice thickness of 2 mm sema quantification using a within-patients study design. with FBP, hybrid IR (HIR; iDose level 4, Philips Healthcare, The secondary aim was to investigate whether adapting Best, The Netherlands) and model-based IR (MIR; IMR CT densitometry thresholds is a valid way to correct for level 2, Philips Healthcare, Best, The Netherlands). Kernel over- or underestimation at reduced dose and with IR. filter C was used for both FBP and HIR. MIR is a more ad- vanced reconstruction technique with different kernels; Methods therefore, the vendor-recommended kernel filter Body Patients Routine was used for MIR. The volume CT dose index This prospective study was approved by the local Institu- (CTDI ) and dose-length product (DLP) of each acquisi- vol tional Review Board (NL46146.041.13) and all study partici- tion was recorded. The effective dose was calculated by pants provided written informed consent. Patients aged multiplying the DLP with a conversion factor of 0.0144 ≥ 50 years scheduled for follow-up of ≥ 1 known small (100 kVp) or 0.0145 (120 kVp) [19]. pulmonary nodules were eligible for inclusion. The in- fluence of dose reduction and IR on pulmonary nodule Emphysema quantification volume and computer-aided detection of pulmonary Semi-automatic commercially available software (Intelli- nodules was previously investigated in the same study Space version 8, COPD tool, Philips Healthcare, Best, population [17, 18]. The Netherlands) was used for emphysema quantifica- tion. The noise reduction option in the software was not Image acquisition used. The software segments airways first, followed by Image acquisition was performed on a 256-slice CT system the lungs and finally the different lobes. No manual (Brilliance iCT; Philips Healthcare, Best, The Netherlands). segmentation was needed. Subsequently, a histogram An unenhanced chest CT was acquired during inspiration. (attenuation curve) is made which displays the number The routine dose acquisition was performed with a tube of voxels with a certain density (Fig. 1). Emphysema voltageof100 kVp(body weight<80kg) or 120kV(body can be quantified by using either a percentile of the Table 1 Percentage of emphysema using the -950 HU threshold and the perc method at different dose levels reconstructed with FBP, HIR and MIR -950 HU (%) ICC (95% CI) Perc (HU-value) ICC (95%CI) Routine dose FBP 5.1 (1.7–8.4) NA -923 (-936 – -895) NA a a HIR 1.5 (0.1–4.5) 0.63 (0.00–0.88) -914 (-927 – -881) 0.91 (0.00–0.98) a a MIR 0.9 (0.0–3.6) 0.50 (0.00–0.80) -913 (-926 – -879) 0.88 (0.01–0.97) 45% reduced dose a a FBP 8.0 (3.3–12.4) 0.83 (0.00–0.96) -932 (-944 – -898) 0.93 (0.39–0.98) a a HIR 2.5 (0.2–5.1) 0.79 (0.01–0.94) -916 (-929 – -875) 0.89 (0.28–0.97) a a MIR 1.2 (0.0–3.1) 0.51 (0.00–0.81) -913 (-927 – -867) 0.83 (0.07–0.95) 60% reduced dose a a FBP 10.2 (5.5–14.7) 0.63 (0.00–0.89) -940 (-949 – -912) 0.77 (0.00–0.94) a a HIR 2.7 (0.5–6.2) 0.83 (0.24–0.95) -917 (-931 – -881) 0.87 (0.62–0.95) a a MIR 1.3 (0.0–3.2) 0.49 (0.00–0.79) -911 (-927 – -874) 0.80 (0.28–0.93) 75% reduced dose a a FBP 14.3 (9.7–19.6) 0.42 (0.00–0.79) -948 (-961 – -925) 0.59 (0.00–0.88) a a HIR 3.5 (0.7–8.1) 0.92 (0.76–0.97) -921 (-935 – -878) 0.94 (0.81–0.98) a a MIR 0.9 (0.0–4.2) 0.47 (0.00–0.78) -914 (-927 – -869) 0.84 (0.16–0.95) Values represent the median (interquartile range). The ICC compares with the reference standard, namely FBP at routine dose Statistically significant difference compared to FBP at routine dose with a Bonferroni corrected p value of 0.017 FBP filtered back projection, HIR hybrid iterative reconstruction, ICC intraclass correlation coefficient, MIR model-based iterative reconstruction den Harder et al. European Radiology Experimental (2018) 2:30 Page 4 of 10 attenuation curve or the percentage of voxels below a with 100 kVp (< 80 kg) and twelve patients (54%) with certain HU value. On the routine dose acquisition re- 120 kVp (≥ 80 kg). The median height of the patients constructed with FBP, emphysema was defined as a HU was 169 cm (163–176 cm) and the median weight was value which describes the lowest 15% of the segmented 83 kg (74–92 kg) resulting in a body mass index of 28.6 2 2 lungs (perc ). Furthermore, the percentage of voxels kg/m (26.0–31.4 kg/m ). The median CTDI was 15 vol with a HU value of -950 HU or lower (percentage 4.1 mGy at routine dose and 2.2, 1.6 and 1.0 mGy at re- emphysema) was calculated. To determine whether the duced dose levels for the 120-kVp acquisitions. For the use of a different percentile or HU threshold is more 100-kVp acquisition, the median CTDI was 2.4, 1.3, vol accurate at reduced dose levels and with IR, additional 1.0 and 0.6 mGy, respectively. The median DLP was measurements were performed as follows (1 percentage 150 (96–169), 84 (53–93), 60 (38–66) and 39 (24– and 10 HU increments): 42) mGy ×cm, respectively, resulting in median effective dose levels of 2.2 (1.4–2.4), 1.2 (0.8–1.3), – reduced dose FBP: perc – perc and -960 HU – 0.9 (0.5–1.0) and 0.6 (0.3–0.6) mSv. 8 35 -1010 HU – HIR:perc – perc and -880 HU – -960 HU Emphysema 1 25 – MIR: perc – perc and -880 HU – -960 HU The percentage of emphysema with FBP at routine dose 1 20 was 5.1% (1.7–8.4%). FBP at reduced dose resulted in a Objective image quality significant overestimation of the percentage of emphy- A region of interest was placed in the ascending aorta at sema, while HIR and MIR resulted in a significant the level of the tracheal bifurcation and in the subcuta- underestimation at all dose levels compared to FBP at neous fat dorsal of the infraspinatus muscle. The noise routine dose (Table 1, Fig. 2). The perc measurements was defined as the standard deviation of the region of resulted in decreased HU values for FBP at reduced interest and the contrast-to-noise ratio (CNR) was calcu- dose, while HIR and MIR resulted in significantly in- lated using the following formula: creased HU values compared to FBP at routine dose (Table 1, Fig. 3). For the -950 HU threshold, HIR at 75% MeanðÞ Aorta −MeanðÞ Fat reduced dose resulted in the highest ICC of 0.92 (0.76– CNR ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 0.97), while the ICC decreased to 0.42 (0.00–0.79) with 2 2 SDðÞ Aorta þ SDðÞ Fat FBP at 75% reduced dose. Overall, the ICC was better with the perc method, resulting in a minimum ICC of 0.59 (0.00–0.88) with FBP at 75% reduced dose and a Statistics maximum ICC of 0.94 (0.81–0.98) with HIR at 75% re- Statistical analysis was performed using SPSS version 21 duced dose. (SPSS Inc., Chicago, IL, USA). The routine dose acquisi- The effect of using different HU thresholds or percen- tion reconstructed with FBP was used as the reference tiles is shown in Fig. 4. Data for each threshold are pro- standard. The Friedman test was used to compare the vided in the supplemental files (Additional file 1:Table reconstructions at each dose level to FBP and post-hoc S1–S4). With FBP at reduced dose, using a lower thresh- analyses were performed with the Wilcoxon signed rank old of -960 HU, -970 HU and -980 HU at 45%, 60% and test. A p value < 0.05 was considered significant for the 75% reduced dose, respectively, resulted in a percentage of Friedman test, while a Bonferroni corrected p value of emphysema that was not significantly different from the 0.017 (0.05/3 reconstructions) was used for the Wilcoxon reference standard. For HIR, a threshold of -930 HU (rou- test. The intraclass correlation coefficient (ICC; two-way tine dose) or -940 HU (reduced dose) approximated the mixed, absolute agreement, single measures) was used to percentage emphysema with FBP at routine dose, while compare reduced dose and iterative reconstruction to the this was -930 HU for MIR (all dose levels). Bland-Altman reference standard. For each dose level and recon- plots are provided in Additional file 2:Figure S1 ofthe struction technique, the optimal adapted threshold for supplemental files. The adapted threshold worked well emphysema quantification was determined. The opti- over the whole range of patients for FBP at reduced dose, mal adapted threshold was also compared to the while with HIR and MIR there was a trend towards under- reference standard using Bland–Altman plots. Results estimation in patients with a small emphysema percentage are displayed as median (interquartile range) unless and in patients with a higher percentage of emphysema specified otherwise. there was an overestimation. FBP at reduced dose required a higher percentile of Results 19%, 22% and 26%, respectively, at 45%, 60% and 75% Twenty-two patients were included. Half of the patients reduced dose to achieve results comparable to the refer- (n = 11) were female. Ten patients (46%) were scanned ence standard. With HIR and MIR, a lower percentile den Harder et al. European Radiology Experimental (2018) 2:30 Page 5 of 10 Fig. 2 Scatterplots of the effect of radiation dose and reconstruction on the percentage emphysema. The y-axis displays the percentage emphysema with FBP at routine dose (reference), while the x-axis displays the percentage emphysema at reduced dose with FPB (a)and with HIR (b) and MIR (c). Values below the diagonal represent an overestimation of the percentage of emphysema as compared to FBP and routine dose, while values above the diagonal represent an underestimation. FBP filtered back projection, HIR hybrid iterative reconstruction, MIR model-based iterative reconstruction den Harder et al. European Radiology Experimental (2018) 2:30 Page 6 of 10 Fig. 3 Scatterplots of the effect of radiation dose and reconstruction on the perc . The y-axis displays the perc with FBP at routine dose 15 15 (reference), while the x-axis displays the perc at reduced dose with FBP (a) and with HIR (b) and MIR (c). Values below the diagonal represent a higher HU value compared to the reference, while values above the diagonal represent a lower HU value compared to the reference. FBP filtered back projection, HIR hybrid iterative reconstruction, MIR model-based iterative reconstruction den Harder et al. European Radiology Experimental (2018) 2:30 Page 7 of 10 Fig. 4 Effect of different thresholds (a) and percentiles (b) on emphysema quantification. For FBP at reduced dose a lower threshold is more appropriate, while with HIR and MIR a higher HU threshold should be used. With the percentile quantification method, FBP at reduced dose requires a higher percentile while with HIR and MIR a lower percentile should be used to achieve the same results as with FBP at routine dose. FBP filtered back projection, HIR hybrid iterative reconstruction, MIR model-based iterative reconstruction was required of 8% (routine dose), 10% (45% reduced standard, while at the lowest dose level noise was dose), 11% (60% reduced dose) or 13% (75% reduced comparable to FBP at routine dose. MIR resulted in a dose) with HIR and 8% (all dose levels) with MIR. significant reduction of noise at all reduced dose Bland–Altman plots are provided in Additional file 2: levels compared to FBP at routine dose. CNR de- Figure S2 of the supplemental files. The adapted thresh- creased with FBP at reduced dose levels. HIR and old worked well over the whole range of patients for all MIR resulted in comparable or improved CNR at all reconstructions. reduced dose levels. Image quality Discussion Noise and CNR are presented in Fig. 5 and Table 2. This study shows the effect of different reconstruction Noise increased with FBP at reduced dose levels, while techniques at four decreasing radiation dose levels. While HIR and MIR resulted in reduced noise. Noise was sig- FBP resulted in an overestimation of emphysema on CT at nificantly lower with HIR at routine dose, 45% reduced reduced dose, both HIR and MIR resulted in an under- dose and 60% reduced dose compared to the reference estimation of the amount of emphysema compared to den Harder et al. European Radiology Experimental (2018) 2:30 Page 8 of 10 Fig. 5 Noise (a and b) and CNR (c) measured at different dose levels with FBP, HIR and MIR. Noise was measured in the aorta (a) and subcutaneous fat (b). The dotted line represents the reference (FBP at routine dose). CNR contrast-to-noise ratio, FBP filtered back projection, HIR hybrid iterative reconstruction, MIR model-based iterative reconstruction reconstruction of the images with FBP at routine dose. with microscopic and macroscopic emphysema findings in Furthermore, we showed that by using different thresh- studies using FBP [8, 9]. The 15th percentile and the -950 olds or percentages in HIR and MIR, it was possible to HU thresholds are widely used; however, different thresh- derive results comparable to FBP at routine dose. olds have been applied in the literature [8, 9, 20, 21]. Several There are two commonly used measures to quantify em- other studies have investigated the effect of dose and image physema on CT based on densitometry, namely the density reconstruction on pulmonary emphysema quantification. at the 15th percentile of the attenuation curve and the per- Schilham et al. [22] compared a clinical routine dose CT ac- centage of voxels below -950 HU. Previous studies indi- quisition with a low dose acquisition in 25 patients. A cated that those thresholds show the strongest correlation post-processing filter was used to reduce the amount of noise in the low dose images and three different thresholds (-950, -930 and -910 HU) were used to quantify emphy- Table 2 Noise and CNR at different dose levels with FBP, HIR sema. The application of the filter resulted in a reduction of and MIR the effect of noise on the emphysema percentage. A differ- Noise (aorta) Noise (fat) CNR ent study by Mets et al. [23] in 75 patients who underwent Routine dose a routine dose CT acquisition reconstructed with FBP and FBP 37.5 (28.1–42.8) 30.2 (25.4–52.0) 4.8 (3.0–5.5) HIR reported an underestimation of the amount of emphy- a a a HIR 18.4 (15.0–22.2) 18.2 (14.6–23.3) 8.3 (6.7–10.1) sema with HIR when the cut-off was not adjusted. In a a a a studybyNishioetal. [24], the application of IR at reduced MIR 11.4 (9.5–12.5) 11.5 (9.8–15.2) 13.0 (10.7–15.9) dose improved the agreement in emphysema quantification 45% reduced dose with routine dose FBP. Three studies comparing a routine a a a FBP 46.6 (38.9–58.4) 39.3 (31.6–78.9) 3.3 (2.1–4.5) dose acquisition with a low-dose acquisition in the same a a a HIR 23.2 (20.5–26.5) 23.5 (17.5–27.8) 6.7 (5.5–8.1) patient all reported an overestimation with low-dose FBP a a a MIR 12.3 (11.3–14.4) 13.5 (10.7–17.2) 11.4 (9.0–13.7) while IR resulted in an underestimation [25–27]. Messerli 60% reduced dose et al. [27] reduced the radiation dose to chest x-ray equiva- a a a lent dose levels of 0.14 mSv; at this dose level, HIR resulted FBP 60.2 (47.8–81.4) 50.8 (42.5–94.6) 2.5 (1.7–3.5) a a a in a similar emphysema measurement as FBP at routine HIR 26.4 (22.9–31.8) 28.1 (21.4–34.7) 5.5 (4.4–6.8) dose (1.7 mSv). Similar results were found in the study by a a a MIR 13.7 (12.2–16.5) 13.9 (12.7–17.2) 11.4 (8.9–12.4) Nishio et al. [28]. Therefore, by carefully selecting the dose 75% reduced dose reduction level, emphysema overestimation can be com- a a a FBP 80.0 (61.4–108.5) 62.8 (47.5–142.6) 2.3 (1.2–2.7) pensated for by using IR, since the latter results in reduced HIR 32.6 (28.5–37.8) 30.4 (24.7–39.6) 4.7 (3.7–5.7) emphysema with CT quantification. To our best know- a a a ledge, only the study by Choo et al. [15] investigated the ef- MIR 16.4 (14.1–19.7) 15.1 (13.8–19.1) 9.2 (7.9–11.3) fects of both HIR and MIR. No dose reduction was used Values are presented as median (interquartile range) Statistically significant difference compared to FBP at routine dose with a and they reported that MIR resulted in a larger underesti- Bonferroni corrected p value of 0.017 mation than HIR compared to FBP, which is comparable to CNR contrast-to-noise ratio, FBP filtered back projection, HIR hybrid iterative reconstruction, MIR model-based iterative reconstruction the results of the current study. den Harder et al. European Radiology Experimental (2018) 2:30 Page 9 of 10 The effect of reconstruction technique and radiation quantification. Values represent the median [interquartile range] HU dose can be explained by the density histogram. IR algo- value at each dose level with HIR. HIR hybrid iterative reconstruction; NA not applicable. Table S4. Effect of different percentiles on rithms result in a different density distribution, which emphysema quantification. Values represent the median [interquartile subsequently affects emphysema quantification. Due to range] HU value at each dose level with MIR. MIR model-based the noise reduction with IR, the extremes of the attenu- iterative reconstruction. (DOCX 35 kb) ation distribution are affected [23], leading to a smaller Additional file 2: Figure S1. Bland–Altman plots for the differences in percentage emphysema when comparing the optimal adapted threshold peak in the density histogram. Dose reduction, on the at each dose level to FBP at routine dose using a − 950 HU threshold. other hand, results in increased image noise, leading to a The continuous line represents the mean difference to the reference broadening of the density histogram [29]. standard while the dotted lines represent the upper and lower limits of agreement (95% limits of agreement). FBP filtered back projection, HIR In the current study, FBP at routine dose was used as hybrid iterative reconstruction; MIR model-based iterative reconstruction. the reference standard. However, ideally a pathological ref- Figure S2. Bland–Altman plots for the differences in HU value when erence standard should be applied or a realistic phantom comparing the optimal adapted threshold at each dose level with FBP at routine dose using the perc15 method. The continuous line represents to determine what is closest to the truth and if thresholds the mean difference to the reference standard while the dotted lines should be adapted. It is important to be aware that differ- represent the upper and lower limits of agreement (95% limits of ences in emphysema quantification can occur and to keep agreement). FBP filtered back projection; HIR hybrid iterative reconstruction; MIR model-based iterative reconstruction. (PDF 578 kb) the radiation dose and reconstruction algorithm constant in longitudinal follow-up studies. Abbreviation Although this within-patient study systematically assessed CNR: Contrast-to-noise ratio; COPD: Obstructive pulmonary disease; the effect of dose and reconstruction on emphysema quan- CT: Computed tomography; CTDIvol: Volume CT dose index; DLP: Dose- tification, there are several limitations. First, the patients length product; FBP: Filtered back projection; HIR: Hybrid IR; ICC: Intraclass correlation coefficient; IR: Iterative reconstruction; MIR: Model-based IR included in this study had a low amount of emphysema. Second, the sample size was relatively low; however, due to Availability of data and materials the within-patients study design, the statistical power of the The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. study was increased. Although we showed that adapting the commonly used thresholds can prevent underestima- Funding tion of emphysema withIR, thesamplesizewas toosmall The authors state that this work has not received any funding. to give a clear recommendation about the optimal thresh- Acknowledgements old. Third, one software package and IR algorithms from The authors would like to thank Ilse van Rein and Sylvia van der Werf for only one vendor were studied and results may differ for their help with the data collection. other packages and other vendors. Fourth, only an inspira- Authors’ contributions tory chest CT was acquired; therefore, air-trapping could Data collection: AH, SL. Data analysis and interpretation: AH, SL, EB. Drafting not be studied. Fifth, the effect of slice thickness and the article: AH, SL, EB. All authors read and approved the final manuscript. reconstruction kernel were not investigated in the current Ethics approval and consent to participate article. Gierada et al. [30] investigated the effect of recon- Institutional Review Board approval was obtained. Written informed consent struction kernel and slice thickness and reported that pa- was obtained from all subjects (patients) in this study. tients with 10–30% emphysema are most sensitive for the Consent for publication effect of kernel and slice thickness, while lower emphy- All authors provided consent for publication. sema percentages (such as in the current study) resulted in more stable measurements. Competing interests Julien Milles is an employee of Philips Healthcare. All other authors declare In conclusion, as compared to FBP at routine dose, that they have no competing interests. both HIR and MIR result in an underestimation of CT emphysema at routine dose and reduced dose while FBP Publisher’sNote results in an overestimation at reduced dose. This can Springer Nature remains neutral 12with regard to jurisdictional claims in potentially be solved by using adapted thresholds. published maps and institutional affiliations. Author details Additional files Department of Radiology, University Medical Center Utrecht, Utrecht, The Netherlands. Department of Radiology, Isala hospital, Zwolle, The 3 4 Netherlands. Philips Healthcare, Best, The Netherlands. Department of Additional file 1: Table S1. Effect of different HU thresholds on Radiology, Erasmus Medical Center, Rotterdam, The Netherlands. emphysema quantification. Values represent the median [interquartile range] percentage emphysema at each dose level with FBP, HIR and MIR. FBP filtered Received: 5 June 2018 Accepted: 6 August 2018 back projection, HIR hybrid iterative reconstruction; MIR model-based iterative reconstruction, NA not applicable. Table S2. Effect of different percentiles on emphysema quantification. Values represent the median [interquartile range] References HU value at each dose level with FBP. FBP filtered back projection; NA 1. Prosch H (2014) Implementation of lung cancer screening: promises and not applicable. Table S3. Effect of different percentiles on emphysema hurdles. Transl Lung Cancer Res 3:286–290 den Harder et al. European Radiology Experimental (2018) 2:30 Page 10 of 10 2. Pompe E, Galbán CJ, Ross BD et al (2017) Parametric response mapping on 24. Nishio M, Koyama H, Ohno Y et al (2016) Emphysema quantification using chest computed tomography associates with clinical and functional ultralow-dose CT with iterative reconstruction and filtered back projection. parameters in chronic obstructive pulmonary disease. Respir Med 123:48–55 AJR Am J Roentgenol 206:1184–1192 3. Mets OM, Schmidt M, Buckens CF et al (2013) Diagnosis of chronic obstructive 25. Wang R, Sui X, Schoepf UJ et al (2015) Ultralow-radiation-dose chest CT: accuracy pulmonary disease in lung cancer screening computed tomography scans: for lung densitometry and emphysema detection. AJR Am J Roentgenol 204: independent contribution of emphysema, air trapping and bronchial wall 743–749 thickening. Respir Res 14:59 26. Nishio M, Matsumoto S, Ohno Y et al (2012) Emphysema quantification by low-dose CT: potential impact of adaptive iterative dose reduction using 3D 4. Madani A, Van Muylem A, de Maertelaer V, Zanen J, Gevenois PA (2008) Pulmonary processing. AJR Am J Roentgenol 199:595–601 emphysema: size distribution of emphysematous spaces on multidetector CT 27. Messerli M, Ottilinger T, Warschkow R et al (2017) Emphysema images--comparison with macroscopic and microscopic morphometry. Radiology quantification and lung volumetry in chest X-ray equivalent ultralow dose 248:1036–1041 CT - intra-individual comparison to standard dose CT. Eur J Radiol 91:1–9 5. Madani A, Zanen J, de Maertelaer V, Gevenois PA (2006) Pulmonary 28. Nishio M, Matsumoto S, Seki S et al (2014) Emphysema quantification on emphysema: objective quantification at multi-detector row CT—comparison low-dose CT using percentage of low-attenuation volume and size with macroscopic and microscopic morphometry. Radiology 238:1036–1043 distribution of low-attenuation lung regions: effects of adaptive iterative 6. Mohamed Hoesein FA, de Jong PA (2016) Landmark papers in respiratory dose reduction using 3D processing. Eur J Radiol 83:2268–2276 medicine: automatic quantification of emphysema and airways disease on 29. Yuan R, Mayo JR, Hogg JC et al (2007) The effects of radiation dose and CT computed tomography. Breathe (Sheff) 12:79–81 manufacturer on measurements of lung densitometry. Chest 132:617–623 7. Mohamed Hoesein FA, de Hoop B, Zanen P et al (2011) CT-quantified 30. Gierada DS, Bierhals AJ, Choong CK et al (2010) Effects of CT section thickness emphysema in male heavy smokers: association with lung function decline. and reconstruction kernel on emphysema quantification relationship to the Thorax 66:782–787 magnitude of the CT emphysema index. Acad Radiol 17:146–156 8. Ostridge K, Wilkinson TM (2016) Present and future utility of computed tomography scanning in the assessment and management of COPD. Eur Respir J 48:216–228 9. Bankier AA, De Maertelaer V, Keyzer C, Gevenois PA (1999) Pulmonary emphysema: subjective visual grading versus objective quantification with macroscopic morphometry and thin-section CT densitometry. Radiology 211:851–858 10. Pistolesi M (2009) Beyond airflow limitation: another look at COPD. Thorax 64:2–4 11. Willemink MJ, de Jong PA, Leiner T et al (2013) Iterative reconstruction techniques for computed tomography part 1: technical principles. Eur Radiol 23:1623–1631 12. Willemink MJ, Leiner T, de Jong PA et al (2013) Iterative reconstruction techniques for computed tomography part 2: initial results in dose reduction and image quality. Eur Radiol 23:1632–1642 13. den Harder AM, Willemink MJ, de Ruiter QM et al (2015) Achievable dose reduction using iterative reconstruction for chest computed tomography: a systematic review. Eur J Radiol 84:2307–2713 14. Baumueller S, Winklehner A, Karlo C et al (2012) Low-dose CT of the lung: potential value of iterative reconstructions. Eur Radiol 22:2597–2606 15. Choo JY,Goo JM, Lee CH,ParkCM, Park SJ,ShimMS(2014)Quantitative analysis of emphysema and airway measurements according to iterative reconstruction algorithms: comparison of filtered back projection, adaptive statistical iterative reconstruction and model-based iterative reconstruction. Eur Radiol 24:799–806 16. Neroladaki A, Botsikas D, Boudabbous S, Becker CD, Montet X (2013) Computed tomography of the chest with model-based iterative reconstruction using a radiation exposure similar to chest X-ray examination: preliminary observations. Eur Radiol 23:360–366 17. den Harder AM, Willemink MJ, van Hamersvelt RW et al (2016) Effect of radiation dose reduction and iterative reconstruction on computer-aided detection of pulmonary nodules: intra-individual comparison. Eur J Radiol 85:346–351 18. den Harder AM, Willemink MJ, van Hamersvelt RW et al (2016) Pulmonary nodule volumetry at different low computed tomography radiation dose levels with hybrid and model-based iterative reconstruction: a within patient analysis. J Comput Assist Tomogr 40:578–583 19. Deak PD, Smal Y, Kalender WA (2010) Multisection CT protocols: sex- and age-specific conversion factors used to determine effective dose from dose- length product. Radiology 257:158–166 20. Hackx M, Bankier AA, Gevenois PA (2012) Chronic obstructive pulmonary disease: CT quantification of airways disease. Radiology 265:34–48 21. Madani A, De Maertelaer V, Zanen J, Gevenois PA (2007) Pulmonary emphysema: radiation dose and section thickness at multidetector CT quantification--comparison with macroscopic and microscopic morphometry. Radiology 243:250–257 22. Schilham AM, van Ginneken B, Gietema H, Prokop M (2006) Local noise weighted filtering for emphysema scoring of low-dose CT images. IEEE Trans Med Imaging 25:451–463 23. Mets OM, Willemink MJ, de Kort FP et al (2012) The effect of iterative reconstruction on computed tomography assessment of emphysema, air trapping and airway dimensions. Eur Radiol 22:2103–2109 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png European Radiology Experimental Springer Journals

Emphysema quantification using chest CT: influence of radiation dose reduction and reconstruction technique

Download PDF
10 pages

Loading next page...
 
/lp/springer-journals/emphysema-quantification-using-chest-ct-influence-of-radiation-dose-YB4Sv5vhg3

References (30)

Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Medicine & Public Health; Imaging / Radiology; Diagnostic Radiology; Interventional Radiology; Neuroradiology; Ultrasound; Internal Medicine
eISSN
2509-9280
DOI
10.1186/s41747-018-0064-3
Publisher site
See Article on Publisher Site

Abstract

Background: Computed tomography (CT) emphysema quantification is affected by both radiation dose (i.e. image noise) and reconstruction technique. At reduced dose, filtered back projection (FBP) results in an overestimation of the amount of emphysema due to higher noise levels, while the use of iterative reconstruction (IR) can result in an underestimation due to reduced noise. The objective of this study was to determine the influence of dose reduction and hybrid IR (HIR) or model-based IR (MIR) on CT emphysema quantification. Methods: Twenty-two patients underwent inspiratory chest CT scan at routine radiation dose and at 45%, 60% and 75% reduced radiation dose. Acquisitions were reconstructed with FBP, HIR and MIR. Emphysema was quantified using the 15th percentile of the attenuation curve and the percentage of voxels below -950 HU. To determine whether the use of a different percentile or HU threshold is more accurate at reduced dose levels and with IR, additional measurements were performed using different percentiles and HU thresholds to determine the optimal combination. Results: Dose reduction resulted in a significant overestimation of emphysema, while HIR and MIR resulted in an underestimation. Lower HU thresholds with FBP at reduced dose and higher HU thresholds with HIR and MIR resulted in emphysema percentages comparable to the reference. The 15th percentile quantification method showed similar results as the HU threshold method. Conclusions: This within-patients study showed that CT emphysema quantification is significantly affected by dose reduction and IR. This can potentially be solved by adapting commonly used thresholds. Keywords: Densitometry, Emphysema, Radiation dosage, Thorax, Tomography (x-ray computed) Key points was achieved at 75% reduced dose with hybrid iterative reconstruction Dose reduction resulted in a significant CT overestimation of emphysema, while iterative Background reconstruction resulted in a significant Chest computed tomography (CT) offers the possibility of underestimation quantifying the amount of emphysema. The number of This can potentially be solved by adapting the chest CT acquisitions is expected to increase due to the commonly used densitometry thresholds favourable results of the National Lung Screening Trial [1] The maximal intraclass correlation coefficient and the interest in subtyping chronic obstructive pulmon- between reduced dose and the reference standard ary disease (COPD) patients [2]. Additional quantification of emphysema on screening CT acquisitions will therefore likely gain importance. Furthermore, this additional infor- * Correspondence: a.m.denharder@umcutrecht.nl mation may also contribute to optimisation of the benefits Department of Radiology, University Medical Center Utrecht, Utrecht, The and cost-effectiveness of CT screening [3]. CT can be used Netherlands Full list of author information is available at the end of the article to both identify patients with emphysema as well as to © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. den Harder et al. European Radiology Experimental (2018) 2:30 Page 2 of 10 monitor progression in patients with COPD. Although em- vivo information about pathological changes and allows physema is traditionally a pathology-based diagnosis [4, 5], for differentiation between airway obstruction and em- CT densitometry of the lungs has demonstrated it to be as- physematous destruction [10]. sociated with airflow obstruction, forced expiratory volume The increasing number of chest CT scans has urged in 1 s and severity according to the Global initiative for the importance of radiation dose reduction. However, chronic Obstructive Lung Disease (GOLD) criteria [6–8]. dose reduction leads to higher noise levels, especially CT densitometry is based on either the 15th percentile of when images are reconstructed using conventional fil- the attenuation curve or the percentage of voxels below tered back projection (FBP). Therefore, several iterative -950 HU because those parameters show the strongest reconstruction (IR) techniques were developed to reduce correlation with microscopic and macroscopic emphy- image noise [11, 12]. Recent studies showed that the ra- sema findings [8, 9]. Although pulmonary function tests diation dose of unenhanced chest CT can be reduced to measure limitation of airflow, they are not able to dif- sub-millisievert dose levels when IR is applied [13]. ferentiate between airway obstruction and emphysema- It is known that CT emphysema quantification is af- tous destruction. CT, on the other hand, provides in fected by both radiation dose (i.e. image noise) and Fig. 1 Example of the semi-automatic software which was used for emphysema quantification. First, the airways, lungs and lung lobes are segmented (a). Subsequently, a histogram is made which displays the number of voxels with a certain density (b). In this example the percentage of voxels below -950 HU is displayed den Harder et al. European Radiology Experimental (2018) 2:30 Page 3 of 10 reconstruction technique. At reduced dose, FBP results weight ≥ 80 kg). The tube current-time product was 60 in an overestimation of the amount of emphysema due mAs at routine dose and subsequently decreased to 33, 24 to higher noise levels, while the use of IR can result in and 15 mAs to achieve 45%, 60% and 75% dose reduction, an underestimation due to reduced noise [14–16]. respectively. All four acquisitions were acquired consecu- The primary aim of the current study was to determine tively in a single session. Automatic exposure control was the effect of both dose reduction and IR on CT emphy- off. Images were reconstructed at a slice thickness of 2 mm sema quantification using a within-patients study design. with FBP, hybrid IR (HIR; iDose level 4, Philips Healthcare, The secondary aim was to investigate whether adapting Best, The Netherlands) and model-based IR (MIR; IMR CT densitometry thresholds is a valid way to correct for level 2, Philips Healthcare, Best, The Netherlands). Kernel over- or underestimation at reduced dose and with IR. filter C was used for both FBP and HIR. MIR is a more ad- vanced reconstruction technique with different kernels; Methods therefore, the vendor-recommended kernel filter Body Patients Routine was used for MIR. The volume CT dose index This prospective study was approved by the local Institu- (CTDI ) and dose-length product (DLP) of each acquisi- vol tional Review Board (NL46146.041.13) and all study partici- tion was recorded. The effective dose was calculated by pants provided written informed consent. Patients aged multiplying the DLP with a conversion factor of 0.0144 ≥ 50 years scheduled for follow-up of ≥ 1 known small (100 kVp) or 0.0145 (120 kVp) [19]. pulmonary nodules were eligible for inclusion. The in- fluence of dose reduction and IR on pulmonary nodule Emphysema quantification volume and computer-aided detection of pulmonary Semi-automatic commercially available software (Intelli- nodules was previously investigated in the same study Space version 8, COPD tool, Philips Healthcare, Best, population [17, 18]. The Netherlands) was used for emphysema quantifica- tion. The noise reduction option in the software was not Image acquisition used. The software segments airways first, followed by Image acquisition was performed on a 256-slice CT system the lungs and finally the different lobes. No manual (Brilliance iCT; Philips Healthcare, Best, The Netherlands). segmentation was needed. Subsequently, a histogram An unenhanced chest CT was acquired during inspiration. (attenuation curve) is made which displays the number The routine dose acquisition was performed with a tube of voxels with a certain density (Fig. 1). Emphysema voltageof100 kVp(body weight<80kg) or 120kV(body can be quantified by using either a percentile of the Table 1 Percentage of emphysema using the -950 HU threshold and the perc method at different dose levels reconstructed with FBP, HIR and MIR -950 HU (%) ICC (95% CI) Perc (HU-value) ICC (95%CI) Routine dose FBP 5.1 (1.7–8.4) NA -923 (-936 – -895) NA a a HIR 1.5 (0.1–4.5) 0.63 (0.00–0.88) -914 (-927 – -881) 0.91 (0.00–0.98) a a MIR 0.9 (0.0–3.6) 0.50 (0.00–0.80) -913 (-926 – -879) 0.88 (0.01–0.97) 45% reduced dose a a FBP 8.0 (3.3–12.4) 0.83 (0.00–0.96) -932 (-944 – -898) 0.93 (0.39–0.98) a a HIR 2.5 (0.2–5.1) 0.79 (0.01–0.94) -916 (-929 – -875) 0.89 (0.28–0.97) a a MIR 1.2 (0.0–3.1) 0.51 (0.00–0.81) -913 (-927 – -867) 0.83 (0.07–0.95) 60% reduced dose a a FBP 10.2 (5.5–14.7) 0.63 (0.00–0.89) -940 (-949 – -912) 0.77 (0.00–0.94) a a HIR 2.7 (0.5–6.2) 0.83 (0.24–0.95) -917 (-931 – -881) 0.87 (0.62–0.95) a a MIR 1.3 (0.0–3.2) 0.49 (0.00–0.79) -911 (-927 – -874) 0.80 (0.28–0.93) 75% reduced dose a a FBP 14.3 (9.7–19.6) 0.42 (0.00–0.79) -948 (-961 – -925) 0.59 (0.00–0.88) a a HIR 3.5 (0.7–8.1) 0.92 (0.76–0.97) -921 (-935 – -878) 0.94 (0.81–0.98) a a MIR 0.9 (0.0–4.2) 0.47 (0.00–0.78) -914 (-927 – -869) 0.84 (0.16–0.95) Values represent the median (interquartile range). The ICC compares with the reference standard, namely FBP at routine dose Statistically significant difference compared to FBP at routine dose with a Bonferroni corrected p value of 0.017 FBP filtered back projection, HIR hybrid iterative reconstruction, ICC intraclass correlation coefficient, MIR model-based iterative reconstruction den Harder et al. European Radiology Experimental (2018) 2:30 Page 4 of 10 attenuation curve or the percentage of voxels below a with 100 kVp (< 80 kg) and twelve patients (54%) with certain HU value. On the routine dose acquisition re- 120 kVp (≥ 80 kg). The median height of the patients constructed with FBP, emphysema was defined as a HU was 169 cm (163–176 cm) and the median weight was value which describes the lowest 15% of the segmented 83 kg (74–92 kg) resulting in a body mass index of 28.6 2 2 lungs (perc ). Furthermore, the percentage of voxels kg/m (26.0–31.4 kg/m ). The median CTDI was 15 vol with a HU value of -950 HU or lower (percentage 4.1 mGy at routine dose and 2.2, 1.6 and 1.0 mGy at re- emphysema) was calculated. To determine whether the duced dose levels for the 120-kVp acquisitions. For the use of a different percentile or HU threshold is more 100-kVp acquisition, the median CTDI was 2.4, 1.3, vol accurate at reduced dose levels and with IR, additional 1.0 and 0.6 mGy, respectively. The median DLP was measurements were performed as follows (1 percentage 150 (96–169), 84 (53–93), 60 (38–66) and 39 (24– and 10 HU increments): 42) mGy ×cm, respectively, resulting in median effective dose levels of 2.2 (1.4–2.4), 1.2 (0.8–1.3), – reduced dose FBP: perc – perc and -960 HU – 0.9 (0.5–1.0) and 0.6 (0.3–0.6) mSv. 8 35 -1010 HU – HIR:perc – perc and -880 HU – -960 HU Emphysema 1 25 – MIR: perc – perc and -880 HU – -960 HU The percentage of emphysema with FBP at routine dose 1 20 was 5.1% (1.7–8.4%). FBP at reduced dose resulted in a Objective image quality significant overestimation of the percentage of emphy- A region of interest was placed in the ascending aorta at sema, while HIR and MIR resulted in a significant the level of the tracheal bifurcation and in the subcuta- underestimation at all dose levels compared to FBP at neous fat dorsal of the infraspinatus muscle. The noise routine dose (Table 1, Fig. 2). The perc measurements was defined as the standard deviation of the region of resulted in decreased HU values for FBP at reduced interest and the contrast-to-noise ratio (CNR) was calcu- dose, while HIR and MIR resulted in significantly in- lated using the following formula: creased HU values compared to FBP at routine dose (Table 1, Fig. 3). For the -950 HU threshold, HIR at 75% MeanðÞ Aorta −MeanðÞ Fat reduced dose resulted in the highest ICC of 0.92 (0.76– CNR ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 0.97), while the ICC decreased to 0.42 (0.00–0.79) with 2 2 SDðÞ Aorta þ SDðÞ Fat FBP at 75% reduced dose. Overall, the ICC was better with the perc method, resulting in a minimum ICC of 0.59 (0.00–0.88) with FBP at 75% reduced dose and a Statistics maximum ICC of 0.94 (0.81–0.98) with HIR at 75% re- Statistical analysis was performed using SPSS version 21 duced dose. (SPSS Inc., Chicago, IL, USA). The routine dose acquisi- The effect of using different HU thresholds or percen- tion reconstructed with FBP was used as the reference tiles is shown in Fig. 4. Data for each threshold are pro- standard. The Friedman test was used to compare the vided in the supplemental files (Additional file 1:Table reconstructions at each dose level to FBP and post-hoc S1–S4). With FBP at reduced dose, using a lower thresh- analyses were performed with the Wilcoxon signed rank old of -960 HU, -970 HU and -980 HU at 45%, 60% and test. A p value < 0.05 was considered significant for the 75% reduced dose, respectively, resulted in a percentage of Friedman test, while a Bonferroni corrected p value of emphysema that was not significantly different from the 0.017 (0.05/3 reconstructions) was used for the Wilcoxon reference standard. For HIR, a threshold of -930 HU (rou- test. The intraclass correlation coefficient (ICC; two-way tine dose) or -940 HU (reduced dose) approximated the mixed, absolute agreement, single measures) was used to percentage emphysema with FBP at routine dose, while compare reduced dose and iterative reconstruction to the this was -930 HU for MIR (all dose levels). Bland-Altman reference standard. For each dose level and recon- plots are provided in Additional file 2:Figure S1 ofthe struction technique, the optimal adapted threshold for supplemental files. The adapted threshold worked well emphysema quantification was determined. The opti- over the whole range of patients for FBP at reduced dose, mal adapted threshold was also compared to the while with HIR and MIR there was a trend towards under- reference standard using Bland–Altman plots. Results estimation in patients with a small emphysema percentage are displayed as median (interquartile range) unless and in patients with a higher percentage of emphysema specified otherwise. there was an overestimation. FBP at reduced dose required a higher percentile of Results 19%, 22% and 26%, respectively, at 45%, 60% and 75% Twenty-two patients were included. Half of the patients reduced dose to achieve results comparable to the refer- (n = 11) were female. Ten patients (46%) were scanned ence standard. With HIR and MIR, a lower percentile den Harder et al. European Radiology Experimental (2018) 2:30 Page 5 of 10 Fig. 2 Scatterplots of the effect of radiation dose and reconstruction on the percentage emphysema. The y-axis displays the percentage emphysema with FBP at routine dose (reference), while the x-axis displays the percentage emphysema at reduced dose with FPB (a)and with HIR (b) and MIR (c). Values below the diagonal represent an overestimation of the percentage of emphysema as compared to FBP and routine dose, while values above the diagonal represent an underestimation. FBP filtered back projection, HIR hybrid iterative reconstruction, MIR model-based iterative reconstruction den Harder et al. European Radiology Experimental (2018) 2:30 Page 6 of 10 Fig. 3 Scatterplots of the effect of radiation dose and reconstruction on the perc . The y-axis displays the perc with FBP at routine dose 15 15 (reference), while the x-axis displays the perc at reduced dose with FBP (a) and with HIR (b) and MIR (c). Values below the diagonal represent a higher HU value compared to the reference, while values above the diagonal represent a lower HU value compared to the reference. FBP filtered back projection, HIR hybrid iterative reconstruction, MIR model-based iterative reconstruction den Harder et al. European Radiology Experimental (2018) 2:30 Page 7 of 10 Fig. 4 Effect of different thresholds (a) and percentiles (b) on emphysema quantification. For FBP at reduced dose a lower threshold is more appropriate, while with HIR and MIR a higher HU threshold should be used. With the percentile quantification method, FBP at reduced dose requires a higher percentile while with HIR and MIR a lower percentile should be used to achieve the same results as with FBP at routine dose. FBP filtered back projection, HIR hybrid iterative reconstruction, MIR model-based iterative reconstruction was required of 8% (routine dose), 10% (45% reduced standard, while at the lowest dose level noise was dose), 11% (60% reduced dose) or 13% (75% reduced comparable to FBP at routine dose. MIR resulted in a dose) with HIR and 8% (all dose levels) with MIR. significant reduction of noise at all reduced dose Bland–Altman plots are provided in Additional file 2: levels compared to FBP at routine dose. CNR de- Figure S2 of the supplemental files. The adapted thresh- creased with FBP at reduced dose levels. HIR and old worked well over the whole range of patients for all MIR resulted in comparable or improved CNR at all reconstructions. reduced dose levels. Image quality Discussion Noise and CNR are presented in Fig. 5 and Table 2. This study shows the effect of different reconstruction Noise increased with FBP at reduced dose levels, while techniques at four decreasing radiation dose levels. While HIR and MIR resulted in reduced noise. Noise was sig- FBP resulted in an overestimation of emphysema on CT at nificantly lower with HIR at routine dose, 45% reduced reduced dose, both HIR and MIR resulted in an under- dose and 60% reduced dose compared to the reference estimation of the amount of emphysema compared to den Harder et al. European Radiology Experimental (2018) 2:30 Page 8 of 10 Fig. 5 Noise (a and b) and CNR (c) measured at different dose levels with FBP, HIR and MIR. Noise was measured in the aorta (a) and subcutaneous fat (b). The dotted line represents the reference (FBP at routine dose). CNR contrast-to-noise ratio, FBP filtered back projection, HIR hybrid iterative reconstruction, MIR model-based iterative reconstruction reconstruction of the images with FBP at routine dose. with microscopic and macroscopic emphysema findings in Furthermore, we showed that by using different thresh- studies using FBP [8, 9]. The 15th percentile and the -950 olds or percentages in HIR and MIR, it was possible to HU thresholds are widely used; however, different thresh- derive results comparable to FBP at routine dose. olds have been applied in the literature [8, 9, 20, 21]. Several There are two commonly used measures to quantify em- other studies have investigated the effect of dose and image physema on CT based on densitometry, namely the density reconstruction on pulmonary emphysema quantification. at the 15th percentile of the attenuation curve and the per- Schilham et al. [22] compared a clinical routine dose CT ac- centage of voxels below -950 HU. Previous studies indi- quisition with a low dose acquisition in 25 patients. A cated that those thresholds show the strongest correlation post-processing filter was used to reduce the amount of noise in the low dose images and three different thresholds (-950, -930 and -910 HU) were used to quantify emphy- Table 2 Noise and CNR at different dose levels with FBP, HIR sema. The application of the filter resulted in a reduction of and MIR the effect of noise on the emphysema percentage. A differ- Noise (aorta) Noise (fat) CNR ent study by Mets et al. [23] in 75 patients who underwent Routine dose a routine dose CT acquisition reconstructed with FBP and FBP 37.5 (28.1–42.8) 30.2 (25.4–52.0) 4.8 (3.0–5.5) HIR reported an underestimation of the amount of emphy- a a a HIR 18.4 (15.0–22.2) 18.2 (14.6–23.3) 8.3 (6.7–10.1) sema with HIR when the cut-off was not adjusted. In a a a a studybyNishioetal. [24], the application of IR at reduced MIR 11.4 (9.5–12.5) 11.5 (9.8–15.2) 13.0 (10.7–15.9) dose improved the agreement in emphysema quantification 45% reduced dose with routine dose FBP. Three studies comparing a routine a a a FBP 46.6 (38.9–58.4) 39.3 (31.6–78.9) 3.3 (2.1–4.5) dose acquisition with a low-dose acquisition in the same a a a HIR 23.2 (20.5–26.5) 23.5 (17.5–27.8) 6.7 (5.5–8.1) patient all reported an overestimation with low-dose FBP a a a MIR 12.3 (11.3–14.4) 13.5 (10.7–17.2) 11.4 (9.0–13.7) while IR resulted in an underestimation [25–27]. Messerli 60% reduced dose et al. [27] reduced the radiation dose to chest x-ray equiva- a a a lent dose levels of 0.14 mSv; at this dose level, HIR resulted FBP 60.2 (47.8–81.4) 50.8 (42.5–94.6) 2.5 (1.7–3.5) a a a in a similar emphysema measurement as FBP at routine HIR 26.4 (22.9–31.8) 28.1 (21.4–34.7) 5.5 (4.4–6.8) dose (1.7 mSv). Similar results were found in the study by a a a MIR 13.7 (12.2–16.5) 13.9 (12.7–17.2) 11.4 (8.9–12.4) Nishio et al. [28]. Therefore, by carefully selecting the dose 75% reduced dose reduction level, emphysema overestimation can be com- a a a FBP 80.0 (61.4–108.5) 62.8 (47.5–142.6) 2.3 (1.2–2.7) pensated for by using IR, since the latter results in reduced HIR 32.6 (28.5–37.8) 30.4 (24.7–39.6) 4.7 (3.7–5.7) emphysema with CT quantification. To our best know- a a a ledge, only the study by Choo et al. [15] investigated the ef- MIR 16.4 (14.1–19.7) 15.1 (13.8–19.1) 9.2 (7.9–11.3) fects of both HIR and MIR. No dose reduction was used Values are presented as median (interquartile range) Statistically significant difference compared to FBP at routine dose with a and they reported that MIR resulted in a larger underesti- Bonferroni corrected p value of 0.017 mation than HIR compared to FBP, which is comparable to CNR contrast-to-noise ratio, FBP filtered back projection, HIR hybrid iterative reconstruction, MIR model-based iterative reconstruction the results of the current study. den Harder et al. European Radiology Experimental (2018) 2:30 Page 9 of 10 The effect of reconstruction technique and radiation quantification. Values represent the median [interquartile range] HU dose can be explained by the density histogram. IR algo- value at each dose level with HIR. HIR hybrid iterative reconstruction; NA not applicable. Table S4. Effect of different percentiles on rithms result in a different density distribution, which emphysema quantification. Values represent the median [interquartile subsequently affects emphysema quantification. Due to range] HU value at each dose level with MIR. MIR model-based the noise reduction with IR, the extremes of the attenu- iterative reconstruction. (DOCX 35 kb) ation distribution are affected [23], leading to a smaller Additional file 2: Figure S1. Bland–Altman plots for the differences in percentage emphysema when comparing the optimal adapted threshold peak in the density histogram. Dose reduction, on the at each dose level to FBP at routine dose using a − 950 HU threshold. other hand, results in increased image noise, leading to a The continuous line represents the mean difference to the reference broadening of the density histogram [29]. standard while the dotted lines represent the upper and lower limits of agreement (95% limits of agreement). FBP filtered back projection, HIR In the current study, FBP at routine dose was used as hybrid iterative reconstruction; MIR model-based iterative reconstruction. the reference standard. However, ideally a pathological ref- Figure S2. Bland–Altman plots for the differences in HU value when erence standard should be applied or a realistic phantom comparing the optimal adapted threshold at each dose level with FBP at routine dose using the perc15 method. The continuous line represents to determine what is closest to the truth and if thresholds the mean difference to the reference standard while the dotted lines should be adapted. It is important to be aware that differ- represent the upper and lower limits of agreement (95% limits of ences in emphysema quantification can occur and to keep agreement). FBP filtered back projection; HIR hybrid iterative reconstruction; MIR model-based iterative reconstruction. (PDF 578 kb) the radiation dose and reconstruction algorithm constant in longitudinal follow-up studies. Abbreviation Although this within-patient study systematically assessed CNR: Contrast-to-noise ratio; COPD: Obstructive pulmonary disease; the effect of dose and reconstruction on emphysema quan- CT: Computed tomography; CTDIvol: Volume CT dose index; DLP: Dose- tification, there are several limitations. First, the patients length product; FBP: Filtered back projection; HIR: Hybrid IR; ICC: Intraclass correlation coefficient; IR: Iterative reconstruction; MIR: Model-based IR included in this study had a low amount of emphysema. Second, the sample size was relatively low; however, due to Availability of data and materials the within-patients study design, the statistical power of the The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. study was increased. Although we showed that adapting the commonly used thresholds can prevent underestima- Funding tion of emphysema withIR, thesamplesizewas toosmall The authors state that this work has not received any funding. to give a clear recommendation about the optimal thresh- Acknowledgements old. Third, one software package and IR algorithms from The authors would like to thank Ilse van Rein and Sylvia van der Werf for only one vendor were studied and results may differ for their help with the data collection. other packages and other vendors. Fourth, only an inspira- Authors’ contributions tory chest CT was acquired; therefore, air-trapping could Data collection: AH, SL. Data analysis and interpretation: AH, SL, EB. Drafting not be studied. Fifth, the effect of slice thickness and the article: AH, SL, EB. All authors read and approved the final manuscript. reconstruction kernel were not investigated in the current Ethics approval and consent to participate article. Gierada et al. [30] investigated the effect of recon- Institutional Review Board approval was obtained. Written informed consent struction kernel and slice thickness and reported that pa- was obtained from all subjects (patients) in this study. tients with 10–30% emphysema are most sensitive for the Consent for publication effect of kernel and slice thickness, while lower emphy- All authors provided consent for publication. sema percentages (such as in the current study) resulted in more stable measurements. Competing interests Julien Milles is an employee of Philips Healthcare. All other authors declare In conclusion, as compared to FBP at routine dose, that they have no competing interests. both HIR and MIR result in an underestimation of CT emphysema at routine dose and reduced dose while FBP Publisher’sNote results in an overestimation at reduced dose. This can Springer Nature remains neutral 12with regard to jurisdictional claims in potentially be solved by using adapted thresholds. published maps and institutional affiliations. Author details Additional files Department of Radiology, University Medical Center Utrecht, Utrecht, The Netherlands. Department of Radiology, Isala hospital, Zwolle, The 3 4 Netherlands. Philips Healthcare, Best, The Netherlands. Department of Additional file 1: Table S1. Effect of different HU thresholds on Radiology, Erasmus Medical Center, Rotterdam, The Netherlands. emphysema quantification. Values represent the median [interquartile range] percentage emphysema at each dose level with FBP, HIR and MIR. FBP filtered Received: 5 June 2018 Accepted: 6 August 2018 back projection, HIR hybrid iterative reconstruction; MIR model-based iterative reconstruction, NA not applicable. Table S2. Effect of different percentiles on emphysema quantification. Values represent the median [interquartile range] References HU value at each dose level with FBP. FBP filtered back projection; NA 1. Prosch H (2014) Implementation of lung cancer screening: promises and not applicable. Table S3. Effect of different percentiles on emphysema hurdles. Transl Lung Cancer Res 3:286–290 den Harder et al. European Radiology Experimental (2018) 2:30 Page 10 of 10 2. Pompe E, Galbán CJ, Ross BD et al (2017) Parametric response mapping on 24. Nishio M, Koyama H, Ohno Y et al (2016) Emphysema quantification using chest computed tomography associates with clinical and functional ultralow-dose CT with iterative reconstruction and filtered back projection. parameters in chronic obstructive pulmonary disease. Respir Med 123:48–55 AJR Am J Roentgenol 206:1184–1192 3. Mets OM, Schmidt M, Buckens CF et al (2013) Diagnosis of chronic obstructive 25. Wang R, Sui X, Schoepf UJ et al (2015) Ultralow-radiation-dose chest CT: accuracy pulmonary disease in lung cancer screening computed tomography scans: for lung densitometry and emphysema detection. AJR Am J Roentgenol 204: independent contribution of emphysema, air trapping and bronchial wall 743–749 thickening. Respir Res 14:59 26. Nishio M, Matsumoto S, Ohno Y et al (2012) Emphysema quantification by low-dose CT: potential impact of adaptive iterative dose reduction using 3D 4. Madani A, Van Muylem A, de Maertelaer V, Zanen J, Gevenois PA (2008) Pulmonary processing. AJR Am J Roentgenol 199:595–601 emphysema: size distribution of emphysematous spaces on multidetector CT 27. Messerli M, Ottilinger T, Warschkow R et al (2017) Emphysema images--comparison with macroscopic and microscopic morphometry. Radiology quantification and lung volumetry in chest X-ray equivalent ultralow dose 248:1036–1041 CT - intra-individual comparison to standard dose CT. Eur J Radiol 91:1–9 5. Madani A, Zanen J, de Maertelaer V, Gevenois PA (2006) Pulmonary 28. Nishio M, Matsumoto S, Seki S et al (2014) Emphysema quantification on emphysema: objective quantification at multi-detector row CT—comparison low-dose CT using percentage of low-attenuation volume and size with macroscopic and microscopic morphometry. Radiology 238:1036–1043 distribution of low-attenuation lung regions: effects of adaptive iterative 6. Mohamed Hoesein FA, de Jong PA (2016) Landmark papers in respiratory dose reduction using 3D processing. Eur J Radiol 83:2268–2276 medicine: automatic quantification of emphysema and airways disease on 29. Yuan R, Mayo JR, Hogg JC et al (2007) The effects of radiation dose and CT computed tomography. Breathe (Sheff) 12:79–81 manufacturer on measurements of lung densitometry. Chest 132:617–623 7. Mohamed Hoesein FA, de Hoop B, Zanen P et al (2011) CT-quantified 30. Gierada DS, Bierhals AJ, Choong CK et al (2010) Effects of CT section thickness emphysema in male heavy smokers: association with lung function decline. and reconstruction kernel on emphysema quantification relationship to the Thorax 66:782–787 magnitude of the CT emphysema index. Acad Radiol 17:146–156 8. Ostridge K, Wilkinson TM (2016) Present and future utility of computed tomography scanning in the assessment and management of COPD. Eur Respir J 48:216–228 9. Bankier AA, De Maertelaer V, Keyzer C, Gevenois PA (1999) Pulmonary emphysema: subjective visual grading versus objective quantification with macroscopic morphometry and thin-section CT densitometry. Radiology 211:851–858 10. Pistolesi M (2009) Beyond airflow limitation: another look at COPD. Thorax 64:2–4 11. Willemink MJ, de Jong PA, Leiner T et al (2013) Iterative reconstruction techniques for computed tomography part 1: technical principles. Eur Radiol 23:1623–1631 12. Willemink MJ, Leiner T, de Jong PA et al (2013) Iterative reconstruction techniques for computed tomography part 2: initial results in dose reduction and image quality. Eur Radiol 23:1632–1642 13. den Harder AM, Willemink MJ, de Ruiter QM et al (2015) Achievable dose reduction using iterative reconstruction for chest computed tomography: a systematic review. Eur J Radiol 84:2307–2713 14. Baumueller S, Winklehner A, Karlo C et al (2012) Low-dose CT of the lung: potential value of iterative reconstructions. Eur Radiol 22:2597–2606 15. Choo JY,Goo JM, Lee CH,ParkCM, Park SJ,ShimMS(2014)Quantitative analysis of emphysema and airway measurements according to iterative reconstruction algorithms: comparison of filtered back projection, adaptive statistical iterative reconstruction and model-based iterative reconstruction. Eur Radiol 24:799–806 16. Neroladaki A, Botsikas D, Boudabbous S, Becker CD, Montet X (2013) Computed tomography of the chest with model-based iterative reconstruction using a radiation exposure similar to chest X-ray examination: preliminary observations. Eur Radiol 23:360–366 17. den Harder AM, Willemink MJ, van Hamersvelt RW et al (2016) Effect of radiation dose reduction and iterative reconstruction on computer-aided detection of pulmonary nodules: intra-individual comparison. Eur J Radiol 85:346–351 18. den Harder AM, Willemink MJ, van Hamersvelt RW et al (2016) Pulmonary nodule volumetry at different low computed tomography radiation dose levels with hybrid and model-based iterative reconstruction: a within patient analysis. J Comput Assist Tomogr 40:578–583 19. Deak PD, Smal Y, Kalender WA (2010) Multisection CT protocols: sex- and age-specific conversion factors used to determine effective dose from dose- length product. Radiology 257:158–166 20. Hackx M, Bankier AA, Gevenois PA (2012) Chronic obstructive pulmonary disease: CT quantification of airways disease. Radiology 265:34–48 21. Madani A, De Maertelaer V, Zanen J, Gevenois PA (2007) Pulmonary emphysema: radiation dose and section thickness at multidetector CT quantification--comparison with macroscopic and microscopic morphometry. Radiology 243:250–257 22. Schilham AM, van Ginneken B, Gietema H, Prokop M (2006) Local noise weighted filtering for emphysema scoring of low-dose CT images. IEEE Trans Med Imaging 25:451–463 23. Mets OM, Willemink MJ, de Kort FP et al (2012) The effect of iterative reconstruction on computed tomography assessment of emphysema, air trapping and airway dimensions. Eur Radiol 22:2103–2109

Journal

European Radiology Experimental – Springer Journals

Published: Nov 7, 2018

There are no references for this article.