Ifd

SSa

b

c

d

ARR2A

KECQDC

1

ptmfe

eMm

lR

0h

European Journal of Radiology 83 (2014) 57– 63

Contents lists available at ScienceDirect

European Journal of Radiology

jo ur nal ho me page: www.elsev ier .com/ locate /e j rad

mproved correlation between CT emphysema quantification and pulmonaryunction test by density correction of volumetric CT data based on air and aorticensity

ong Soo Kima, Joon Beom Seob,∗, Namkug Kimb, Eun Jin Chaeb, Young Kyung Leec, Yeon Mok Ohd,ang Do Leed

Department of Radiology, Chungnam National University Hospital, Chungnam National University School of Medicine, Republic of KoreaDepartment of Radiology, University of Ulsan College of Medicine, Asan Medical Center, Republic of KoreaDepartment of Radiology, Kyung Hee University Hospital at Gangdong, Republic of KoreaDivision of Pulmonology, Department of Internal Medicine, University of Ulsan College of Medicine, Asan Medical Center, Republic of Korea

a r t i c l e i n f o

rticle history:eceived 9 April 2011eceived in revised form1 September 2011ccepted 27 February 2012

eywords:mphysemaomputed tomography scanneruantitative evaluationensitometryomputer software

a b s t r a c t

Objectives: To determine the improvement of emphysema quantification with density correction and todetermine the optimal site to use for air density correction on volumetric computed tomography (CT).Methods: Seventy-eight CT scans of COPD patients (GOLD II–IV, smoking history 39.2 ± 25.3 pack-years)were obtained from several single-vendor 16-MDCT scanners. After density measurement of aorta,tracheal- and external air, volumetric CT density correction was conducted (two reference values: air,−1000 HU/blood, +50 HU). Using in-house software, emphysema index (EI) and mean lung density (MLD)were calculated. Differences in air densities, MLD and EI prior to and after density correction were evalu-ated (paired t-test). Correlation between those parameters and FEV1 and FEV1/FVC were compared (age-and sex adjusted partial correlation analysis).Results: Measured densities (HU) of tracheal- and external air differed significantly (−990 ± 14, −1016 ± 9,P < 0.001). MLD and EI on original CT data, after density correction using tracheal- and external air also dif-

fered significantly (MLD: −874.9 ± 27.6 vs. −882.3 ± 24.9 vs. −860.5 ± 26.6; EI: 16.8 ± 13.4 vs. 21.1 ± 14.5vs. 9.7 ± 10.5, respectively, P < 0.001). The correlation coefficients between CT quantification indices andFEV1, and FEV1/FVC increased after density correction. The tracheal air correction showed better resultsthan the external air correction.Conclusion: Density correction of volumetric CT data can improve correlations of emphysema quantifi-cation and PFT.

. Introduction

In the assessment of emphysema, lung densitometry using com-uted tomography (CT) has proven to be an important tool forhe measurement of emphysema extent [1–8]. Not only it is well

atched with pathology, it also correlates well with pulmonaryunction tests (PFT), and is sensitive in assessing progression ofmphysema [9–17]. However, there are many sources of error and

Abbreviations: CT, computed tomography; EI, emphysema index; FEV1, forcedxpiratory volume in 1 second; FVC, forced vital capacity; HU, Hounsfield unit;DCT, multi-detector computed tomography; MLD, mean lung density; PFT, pul-onary function test; ROI, region of interest.∗ Corresponding author at: Department of Radiology, University of Ulsan Col-

ege of Medicine, Asan Medical Center, 388-1, Poongnap-dong, Songpa-gu, Seoul,epublic of Korea. Tel.: +82 2 3010 4400; fax: +82 2 476 4719.

E-mail addresses: seojb@amc.seoul.kr, joonbeom.seo@gmail.com (J.B. Seo).

720-048X/$ – see front matter © 2012 Elsevier Ireland Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.ejrad.2012.02.021

© 2012 Elsevier Ireland Ltd. All rights reserved.

variability in emphysema assessments conducted via CT [18–20].In terms of technical factors, discrepancies among CT measure-ments taken from different scanners across vendors and versionsare a well-known issue [21]. Additionally, even when using thesame scanner, variations in radiation exposure, reconstructionkernel, and reconstruction thickness are known to generate dis-crepant measurement results [22,23]. Other possible sources oferror when using the same scanner include tube decay and cali-bration. Regarding patient factors, beam hardening and respiratorylevels are matters of concern. Accordingly, lung densitometry withCT requires a high level of reproducibility in data acquisition andstandardization to reduce inter- and intra-observer variability indensitometric assessments of emphysema. Stoel et al. [18] also

reported that they had identified the sources of error in lungdensitometry with CT and developed a correction method. Intheir correction method, they suggested the use of modified cor-rected thresholds according to the measured blood and air density

5 rnal of Radiology 83 (2014) 57– 63

vfmedth[ulwottl

2

2

b(pOsIywh

tCtmCwatpCcIp

2

MTlotrrrnafm

cfi[

8 S.S. Kim et al. / European Jou

alues. In this study, we focused on the intra-scanner variationrom calibration error and beam hardening and designed a modified

ethod based on their correction method. Firstly, we corrected thentire volumetric multi-detector computed tomography (MDCT)ata based on the reference density value of blood and air. Secondly,he emphysema index (EI) was automatically calculated using in-ouse software via a fixed threshold of −950 Hounsfield units (HU)11,12,19]. We hypothesized that the correction of the entire vol-metric CT data set would provide better density correction for

ung air. The principal objectives of this study were to determinehether our new method of density correction of MDCT data based

n air and blood densities would improve emphysema quantifica-ion in terms of its correlation with PFT, and to determine whetherracheal air or external air was better density correction of meanung density (MLD, HU) and EI (%).

. Materials and methods

.1. Clinical subjects

A total of 257 subjects diagnosed with COPD, i.e., post-ronchodilator ratio between forced expiratory volume in 1 secondFEV1) and forced vital capacity (FVC) < 0.7 and more than 10ack-years of smoking history, were identified from the Koreanbstructive Lung Disease (KOLD) cohort for possible inclusion. All

ubjects had CT scans and PFTs acquired on the same day. Thenstitutional Review Boards from each hospital approved the anal-sis of the clinical and imaging data and written informed consentas obtained from all subjects. Subjects were recruited from 11ospitals in South Korea from June 2005 to December 2009.

For the analysis, we selected subjects with CT scans acquired atwo different hospitals in which the same manufacturer and modelT scanners were installed. Additional selection criteria includedhe following: (1) visualized emphysema on CT, (2) no parenchy-

al destruction due to disease other than emphysema, (3) adequateT image quality without motion artifact. Therefore, 179 subjectsere excluded and 78 subjects (68 men and 10 women) with an

verage age of 65.3 ± 8.2 years (range = 45–82 years) were iden-ified for inclusion. The average smoking history was 39.2 ± 25.3ack-years (range = 10–135 pack-years). The Global Initiative forhronic Obstructive Lung Disease (GOLD) subject classificationsonsisted of GOLD stage II (n = 35, 34.1 ± 28.9 pack-years), stageII (n = 33, 41.7 ± 21.5 pack-years), and stage IV (n = 10, 49.1 ± 21.2ack-years).

.2. CT acquisition and data analysis

Volumetric CT scans were obtained at full inspiration using 16DCT scanners (Brilliance 16; Philips Healthcare, Netherlands).

he CT acquisition parameters were as follows: 16 × 0.75 mm col-imation, 133 mA, 140 kVp, pitch of 1, and 0.75 s/rotation. The scalef attenuation coefficients in these CT scanners ranged from −1024o 3072 HU. Subjects were scanned during suspended full inspi-ation in supine position in a craniocaudal direction. No subjectseceived intravenous contrast medium. The acquired data wereeconstructed using reconstruction algorithm B with 1 mm thick-ess and 1 mm increment. The effective dose of CT protocol waspproximately 10 mSv. Scanned CT data were stored in DICOMormat, which is the international standard for interconnecting

edical imaging devices on standard networks.Using in-house software, whole lung images were automati-

ally extracted, and then the attenuation coefficients of each pixelor region of interest (ROI) in tracheal air, external air, and bloodn the proximal descending aorta were measured and calculated24]. Measurement of the ROI in tracheal air was performed in the

Fig. 1. Measurement sites for region of interest (ROI) in tracheal air (a), external air(b), and blood (c).

tracheal lumen at the level of the aortic arch. The ROI site for exter-nal air was positioned 5 cm anterior to the sternum at the carinalevel and the ROI site for blood density was positioned in the prox-imal descending thoracic aorta at the carina level (Fig. 1). All of the

ROIs were the same size (10 mm, nearly half the tracheal diameter)and round. Those measured density values of each voxels in wholevolumetric CT data set were proportionally corrected based on tworeference values (tracheal air at −1000 HU/blood at +50 HU and

S.S. Kim et al. / European Journal of Radiology 83 (2014) 57– 63 59

F valueP etwe

erwiTpaa

Y

HHCiiHp

Y

aaba

2

e

correlation analysis. The difference in measured correlation valuesbetween the original measurements, tracheal air-based correc-tions, and external air corrections were also evaluated (Hotelling’s

Table 1Patient Characteristics and pulmonary function tests in patients with COPD.

Characteristics Value

No. of patients 78Sex ratio (M/F) 68/10Age (y) 65.3 ± 8.2 (45–82)BMI (kg/m2) 23.9 ± 3.5 (14–33.9)FEV1% pred 49.6 ± 16.2 (21.6–79.4)FVC % pred 68.4 ± 17.5 (33.0–111.9)

ig. 2. Density correction of actually measured CT pixel density based on referenceroportional shifting of density values after density correction, assuming linearity b

xternal air at −1000 HU/blood at +50 HU, respectively) and whichesulted in reassignment of the whole CT pixel (Fig. 2a). In otherords, variations in blood and air density values implied changes

n HU values with an intermediate density between blood and air.herefore, the whole CT pixel corrections were performed using aroportional shift of the plot of actual HU values for air and bloodgainst the ideal values (Fig. 2b). The corrected CT pixel density waschieved using the following formula:

= HUe,air − HUe,blood

HUm,air − HUm,bloodX +

(HUe,blood − HUe,air − HUe,blood

HUm,air − HUm,bloodHUm,blood

)

Ue is the expected HU after correction and HUm is the measuredU values on original data. In other words, HUm,air is the measuredT attenuation values in tracheal- or external air and HUm,blood

s the measured CT attenuation values of blood in the descend-ng thoracic aorta. In patients with COPD, HUe,air is −1000 HU andUe,blood is 50 HU. From the original equation, X is the original CTixel density (HU), and Y is the corrected CT pixel density (HU).

= −1000 − 50HUm,air − HUm,blood

X +(

50 − −1000 − 50HUm,air − HUm,blood

HUm,blood

)

The cutoff level between normal lung density and a low-ttenuated area was defined as −950 HU [11,25]. The EI was defineds the relative area of lung tissue in each CT slice that was occupiedy pixels having values below −950 HU. From the CT data, EI (%)nd MLD (HU) were automatically calculated.

.3. Pulmonary function test

Spirometry was conducted according to American Thoracic Soci-ty guidelines (Vmax 22, SensorMedics; PFDX, MedGraphics) [26].

in air (−1000 HU) and blood (50 HU) results in corrected ideal CT pixel density (a).en air and blood, show how to perform volume density correction (b).

The following values were evaluated: forced vital capacity (FVC),forced expiratory volume in 1 second (FEV1), ratio of FEV1 to FVC(FEV1/FVC). All spirometric values were expressed as the percent-ages of predicted values (percentage of predicted).

2.4. Statistical analysis

All statistical analyses were conducted using SPSS v12.1.1 (SPSS;Chicago, IL). The results were expressed as means ± standard devi-ation (SD). The difference between the measured densities oftracheal air and external air was analyzed via paired t-test. Cor-relation between the measured MLD and EI with PFT parameters(FEV1, FEV1/FVC) were evaluated by age- and sex-adjusted partial

FEV1/FVC % 50.5 ± 10.5 (26.2–68.5)

All data are expressed as mean ± SDs.COPD: chronic obstructive pulmonary disease, BMI: body-mass index, FEV1: forcedexpiratory volume in 1 second, FVC: forced expiratory vital capacity.

6 rnal of Radiology 83 (2014) 57– 63

Ts

3

is

3

afa

taauUac

3t

pAmF00avdvaia

4

uqttm

hsCsaHaoiavotf

Fig. 3. Dramatic changes in emphysema index (EI) after volumetric density correc-tion showing the increase in EI from 2.5% to 24.18% after density correction. Colormapped axial and coronal images in COPD patient before (a) and after (b) volume

0 S.S. Kim et al. / European Jou

-square distribution). P-values of less than 0.05 were regarded astatistically significant.

. Results

The baseline characteristics of the 78 COPD patients are shownn Table 1. Subjects were classified as GOLD stage II (n = 35), GOLDtage III (n = 33) or GOLD stage IV (n = 10).

.1. Change in MLD and EI after air- and blood density correction

The median values of density (HU) in tracheal- and external airnd aortic blood were −992.4, −1021.2, and 30.25. A significant dif-erence in air density was noted between the tracheal- and externalir.

The results of changes in MLD and EI of the CT data sets prioro and after density correction using tracheal air, external air, andortic blood density are shown in Table 2. After correction, all MLDnd EI values were significantly different from the uncorrected val-es and even between tracheal- and external air (Fig. 3, P < 0.001).sing tracheal air correction, a reduction in the value of MLD andn increase in EI were shown, however in the case of external airorrection, the results were opposite.

.2. Comparison of density correction effect using betweenracheal air and external air

The changes in correlation coefficients between MLD/EI and PFTarameters prior to and after air correction are shown in Table 3.fter tracheal- and external air correction, there were improve-ents in correlations between the CT quantification indices and

EV1 (MLD, original, tracheal-, outside air correction: 0.32, 0.52,.45; EI: −0.44, −0.65, −0.49, respectively) and FEV1/FVC (MLD:.49, 0.57, 0.53; EI: −0.60, −0.67, −0.53, respectively). The trachealir correction method showed better improvement of correlationalues than external air correction method. Hotelling’s T-squareistribution test showed significant improvement of correlationalues between MLD and FEV1, MLD and FEV1/FVC and EI and FEV1fter tracheal air correction. Figs. 4 and 5 show the improvementsn the scattergram and the relationship between the EI and FEV1nd between the EI and FEV1/FVC after tracheal air correction.

. Discussion

In this study, we demonstrated that density correction of vol-metric CT data based on air and blood improved emphysemauantification showing with improved correlation with PFT andhat tracheal air correction was a more reliable method thanhe external air correction using the newly modified correction

ethod.For the assessment of emphysema, lung densitometry using CT

as been an important tool for the measurement of emphysemaeverity [1–5]. Additionally, several studies have demonstrated thatT lung densitometry is necessary to detect emphysema progres-ion [27]. Furthermore, this technique is also regarded as a usefulssessment and prediction tool for surgical treatment [28–30].owever, the routine clinical usage of CT lung densitometry inssessing emphysema severity remains limited in scope becausef several factors, one of which is a concern regarding variabil-ty of measurements. This measurement variability is not whollyttributable to variations in scan protocols, or to variations in CT

endor or version. Even when the same scan parameters are usedn the same machine, differences in measurement can occur dueo tube decay, failure of standardization and a variety of patientactors including differences in posture, body-mass index (BMI),

correction (CT attenuation value < −1000 HU, red; <−990 HU, orange; <−980 HU,yellow; <−970 HU, yellow green; <−960 HU, green; <−950 HU, blue: not shownhere).

S.S. Kim et al. / European Journal of Radiology 83 (2014) 57– 63 61

Table 2The results of the correction procedure at each side of recalibration using tracheal air and external air in 78 COPD patients.

Parameters Uncorrected Corrected P-value

Tracheal air External air

MLD (HU) −874.9 ± 27.6 (−879.1)a −882.3 ± 24.9 (−884.8)a −860.5 ± 26.6 (−860.1)a All < .001EI (%, −950) 16.8 ± 13.4 (13.3)a 21.1 ± 14.5 (17.0)a 9.7 ± 10.5 (6.3)a All < .001

AC sema i

bvnb

Fv(

data acquisition errors [18]. Therefore, an accurate, objective, andreliable density correction method is clearly required.

ll data are expressed as mean ± SDs.OPD: chronic obstructive pulmonary disease, MLD: mean lung density, EI: emphya Median value.

reathhold level and so on [18,20]. Even though the errors in theolumetric CT data and CT acquisition may be small, it is recog-

ized that not only X-ray tube ageing and replacement but alsoeam hardening effects related to the thoracic ribs may introduce

ig. 4. Correlation between the emphysema index (EI, y-axis) and forced expiratoryolume in 1 second (FEV1, x-axis) at before (a) and after air correction of tracheal-b) and external air (c).

ndex, HU: Hounsfield unit.

Traditionally, density correction using water density is usedfor general CT density calibrations [1]. However, when assessing

Fig. 5. Correlation between the emphysema index (EI, y-axis) and ratio of forcedexpiratory volume in 1 second to force vital capacity (FEV1/FVC, x-axis) at before (a)and after air correction of tracheal- (b) and external air (c).

62 S.S. Kim et al. / European Journal of

Table 3Correlation coefficients between COPD parameters and PFT in 78 COPD patients.

Parameters FEV1 FEV1/FVC

MLD

Original 0.323 3 0.494 7

Tracheal air correction 0.5212a 0.5746 a

External air correction 0.4526 a 0.5347

EI

Original - 0.443 4 - 0.596 1

Tracheal air correction - 0.6456 a - 0.6691

External air correction - 0.493 0 - 0.533 8

COPD: chronic obstructive pulmonary disease, MLD: mean lung density, EI:emphysema index, FEV1: forced expiratory volume in 1 second, FVC: forced expi-ratory vital capacity, PFT: pulmonary function test.All P < 0.05, age, sex adjusted partial correlation.

ebipsfipeaoatatebtHdttw

lraowtsoHvtapvrir

Acknowledgements

aStatistically significant difference in correction values after tracheal- and exter-nal air correction (Hotelling’s T-square distribution), P < 0.05.

mphysema, additional air density corrections are very important,ecause variations in pulmonary air density can have a profound

nfluence on emphysema assessment [19,20]. Kemerink et al. [31]reviously demonstrated that the HU for air varies between CTcanners, and that variability was attenuated when adjustmentsor differences in air correction were carried out. This shows themportance of an intra- and interscanner air density correctionrocedure to adjust for potential errors in air densitometry. In anffort to ameliorate air density variations, Parr et al. [19] devised

method using a modified corrected emphysema threshold basedn the measured density of air and blood from the scanned image,nd demonstrated an improvement in the correlation betweenhe emphysema measurement and PFT. However, this method is

somewhat limited, and presents difficulties in routine usage dueo the need for time calculations of the emphysema threshold invery case. Additionally, the emphysema threshold varies on a case-y-case basis, which makes it difficult to conduct the tests in aimely manner. Therefore, we suggest a total correction of the pixelU density in the whole volumetric CT data using the measuredensity values of air and blood, based on two reference values ofracheal- and external air (−1000 HU) and blood (+50 HU). Addi-ionally, we suggest a fixed threshold of −950 HU. As in Parr’s study,e achieved an improved correlation with PFT.

An issue associated with these air correction methods is theocation or site at which the air is measured. In air density cor-ection, we could select either an internal air such as the trachea orn air correction site external to the body. However, the air densityf HU differs between the tracheal air and external air, and thereas no comparative information available regarding the value of

racheal- and external air correction in the context of MDCT. In thistudy, the mean density (HU) of air in the ROI of the trachea andutside the patient was −990 ± 14 and −1016 ± 9, respectively. TheU values differ significantly between them and the much higheralues in the tracheal air than in the external air may be attributedo the influence of the beam hardening effect of the thorax, as wells the effect of increased tracheal air humidity related to body tem-erature [18]. Our study results demonstrated a reduction in thealue of MLD and an increase in EI in the case of tracheal air cor-

ection and an increase in the value of MLD and a reduction in EIn the case of external air correction and we found stronger cor-elation coefficient values in tracheal air correction. One possible

Radiology 83 (2014) 57– 63

explanation for this involves the common influence of human bodytissue on CT density variations in the lung tissues and tracheal air.Because the trachea is located in the center of the thorax, the pho-ton deprivation and beam hardening effect caused by the chest walltissue and bony thorax on the lung density measurement would beshared, whereas external air is relatively free from those influences.This observation is partially supported by the fact that the tracheaair density is consistently higher than external air density.

We expect that our modified method of volumetric CT densitycorrection based on air and aortic blood densities, and especiallytracheal air density may provide more reproducible, consistent, andimproved emphysema quantification by diminution of error.

Our study had some limitations. First, the assigned ROI loca-tion in the trachea, external air site, and descending aorta mayhave influenced the reproducibility of the correction assessments;we did not make any determinations as to reproducibility. More-over, the location of the trachea measurement at the aortic archlevel was an arbitrary decision. Furthermore, variations in air den-sity were noted within the trachea depending on the level of thethorax due to the beam hardening effect [18]. Three-dimensionalsegmentation of the trachea may alleviate this shortcoming of ourtechnique; however, the positioning of the trachea in the upper halfof the thorax, where the most severe beam hardening effect occurs,imposes a limitation. Second, we did not consider the effects ofhemoglobin levels on blood density corrections. This may be impor-tant because blood is the main constituent in the lung and changesin hemoglobin level could easily affect the lung density measure-ments [18]. Third, the correlation between EI/MIL and PFT is notcomplete; a better gold standard is pathology, but that is not rele-vant to a clinical study. Fourth, we did not consider the influence ofthe inspiration level during CT scanning. Even with the consistentbreathing instructions, the inspiration level affects the quantitativeassessment. Fifth, the radiation exposure in the applied protocol isrelatively high. Gierada et al. [32] reported that low-radiation-dosetechnique has minimal effect on CT quantification of emphysema,although a statistically significant difference in EI is seen belowthe level of −920 HU threshold. We expect that our density correc-tion method will not be also affected by the low-dose-technique,because we made the final correction and reassignment of wholeCT pixel numbers after shifting air and blood densities to −1000and +50 HU and then we obtained the EI using fixed threshold(−950 HU). Sixth, we did not compare our in-house software withpreviously reported methods of commercially available software,especially for the assessment of smoking-related pulmonary func-tional impairment, phenotype characterization, and clinical stageclassification of smoking-related COPD.

In the future, this correction method could be tested again usingCT data from different CT vendors and versions. This will be impor-tant for multicenter studies, as different CT machines are installedat different centers.

In conclusion, our study results demonstrated that densitycorrections of volumetric CT data based on air and aortic blooddensities can improve emphysema quantifications in terms of cor-relation with PFT. The tracheal air correction technique yieldedbetter results than the external air correction.

Conflict of interest

None of the authors have any competing interests.

This study was supported by a grant (A040153) from the KoreanHealth 21 R&D Project, Ministry of Health, Welfare, Republic of

rnal of

KL

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

S.S. Kim et al. / European Jou

orea. The authors thank the members of the Korean Obstructiveung Disease (KOLD) cohort study group.

eferences

[1] Dowson LJ, Guest PJ, Stockley RA. Longitudinal changes in physiological, radio-logical, and health status measurements in alpha(1)-antitrypsin deficiency andfactors associated with decline. American Journal of Respiratory and CriticalCare Medicine 2001;164:1805–9.

[2] Muller NL, Staples CA, Miller RR, Abboud RT. “Density mask”. An objec-tive method to quantitate emphysema using computed tomography. Chest1988;94:782–7.

[3] Gevenois PA, De Vuyst P, Sy M, et al. Pulmonary emphysema: quantitative CTduring expiration. Radiology 1996;199:825–9.

[4] Nishimura K, Murata K, Yamagishi M, et al. Comparison of different computedtomography scanning methods for quantifying emphysema. Journal of ThoracicImaging 1998;13:193–8.

[5] Mishima M, Itoh H, Sakai H, et al. Optimized scanning conditions of high res-olution CT in the follow-up of pulmonary emphysema. Journal of ComputerAssisted Tomography 1999;23:380–4.

[6] Gietema HA, Schilham AM, van Ginneken B, et al. Monitoring of smoking-induced emphysema with CT in a lung cancer screening setting: detection ofreal increase in extent of emphysema. Radiology 2007;244:890–7.

[7] Cavigli E, Camiciottoli G, Diciotti S, et al. Whole-lung densitometry versus visualassessment of emphysema. European Radiology 2009;19:1686–92.

[8] Gietema HA, Muller NL, Fauerbach PV, et al. Quantifying the extent ofemphysema: factors associated with radiologists’ estimations and quantitativeindices of emphysema severity using the ECLIPSE cohort. Academic Radiology2011;18:661–71.

[9] Stolk J, Stoel BC. Lung densitometry to assess progression of emphysema inchronic obstructive pulmonary disease: time to apply in the clinic? AmericanJournal of Respiratory and Critical Care Medicine 2011;183:1578–80.

10] Bankier AA, De Maertelaer V, Keyzer C, Gevenois PA. Pulmonary emphysema:subjective visual grading versus objective quantification with macroscopicmorphometry and thin-section CT densitometry. Radiology 1999;211:851–8.

11] Gevenois PA, de Maertelaer V, De Vuyst P, Zanen J, Yernault JC. Comparison ofcomputed density and macroscopic morphometry in pulmonary emphysema.American Journal of Respiratory and Critical Care Medicine 1995;152:653–7.

12] Gevenois PA, De Vuyst P, de Maertelaer V, et al. Comparison of com-puted density and microscopic morphometry in pulmonary emphysema.American Journal of Respiratory and Critical Care Medicine 1996;154:187–92.

13] Camiciottoli G, Bartolucci M, Maluccio NM, et al. Spirometrically gated high-

resolution CT findings in COPD: lung attenuation vs lung function and dyspneaseverity. Chest 2006;129:558–64.

14] Heremans A, Verschakelen JA, Van fraeyenhoven L, Demedts M. Measurementof lung density by means of quantitative CT scanning. A study of correlationswith pulmonary function tests. Chest 1992;102:805–11.

[

[

Radiology 83 (2014) 57– 63 63

15] Madani A, De Maertelaer V, Zanen J, Gevenois PA. Pulmonary emphysema:radiation dose and section thickness at multidetector CT quantification –comparison with macroscopic and microscopic morphometry. Radiology2007;243:250–7.

16] Parr DG, Sevenoaks M, Deng C, Stoel BC, Stockley RA. Detection of emphy-sema progression in alpha 1-antitrypsin deficiency using CT densitometry;methodological advances. Respiratory Research 2008;9:21.

17] Matsuoka S, Yamashiro T, Washko GR, et al. Quantitative CT assessment ofchronic obstructive pulmonary disease. Radiographics 2010;30:55–66.

18] Stoel BC, Vrooman HA, Stolk J, Reiber JH. Sources of error in lung densitometrywith CT. Investigative Radiology 1999;34:303–9.

19] Parr DG, Stoel BC, Stolk J, Nightingale PG, Stockley RA. Influence of calibra-tion on densitometric studies of emphysema progression using computedtomography. American Journal of Respiratory and Critical Care Medicine2004;170:883–90.

20] Stoel BC, Stolk J. Optimization and standardization of lung densitome-try in the assessment of pulmonary emphysema. Investigative Radiology2004;39:681–8.

21] Bakker ME, Stolk J, Putter H, et al. Variability in densitometric assessment ofpulmonary emphysema with computed tomography. Investigative Radiology2005;40:777–83.

22] Boedeker KL, McNitt-Gray MF, Rogers SR, et al. Emphysema: effect of recon-struction algorithm on CT imaging measures. Radiology 2004;232:295–301.

23] Ley-Zaporozhan J, Ley S, Weinheimer O, et al. Quantitative analysis of emphy-sema in 3D using MDCT: influence of different reconstruction algorithms.European Journal of Radiology 2008;65:228–34.

24] Lee YK, Oh YM, Lee JH, et al. Quantitative assessment of emphysema, air trap-ping, and airway thickening on computed tomography. Lung 2008;186:157–65.

25] Friedman PJ. Imaging studies in emphysema. Proceedings of the American Tho-racic Society 2008;5:494–500.

26] Standardization of spirometry, 1994 update. American Thoracic Society. Amer-ican Journal of Respiratory and Critical Care Medicine 1995;152:1107–36.

27] Newell Jr JD, Hogg JC, Snider GL. Report of a workshop: quantitative com-puted tomography scanning in longitudinal studies of emphysema. EuropeanRespiratory Journal 2004;23:769–75.

28] Bae KT, Slone RM, Gierada DS, Yusen RD, Cooper JD. Patients with emphysema:quantitative CT analysis before and after lung volume reduction surgery. Workin progress. Radiology 1997;203:705–14.

29] Gierada DS, Yusen RD, Pilgram TK, et al. Repeatability of quantitative CT indexesof emphysema in patients evaluated for lung volume reduction surgery. Radi-ology 2001;220:448–54.

30] Gierada DS, Yusen RD, Villanueva IA, et al. Patient selection for lung volumereduction surgery: an objective model based on prior clinical decisions andquantitative CT analysis. Chest 2000;117:991–8.

31] Kemerink GJ, Lamers RJ, Thelissen GR, van Engelshoven JM. Scanner conformityin CT densitometry of the lungs. Radiology 1995;197:749–52.

32] Gierada DS, Pilgram TK, Whiting BR, et al. Comparison of standard- andlow-radiation-dose CT for quantification of emphysema. American Journal ofRoentgenology 2007;188:42–7.