UBC Faculty Research and Publications

Quantification of lung surface area using computed tomography Yuan, Ren; Nagao, Taishi; Paré, Peter D; Hogg, James C; Sin, Don D; Elliott, Mark W; Loy, Leanna; Xing, Li; Kalloger, Steven E; English, John C; Mayo, John R; Coxson, Harvey O Oct 31, 2010

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12931_2010_Article_996.pdf [ 1.38MB ]
JSON: 52383-1.0223620.json
JSON-LD: 52383-1.0223620-ld.json
RDF/XML (Pretty): 52383-1.0223620-rdf.xml
RDF/JSON: 52383-1.0223620-rdf.json
Turtle: 52383-1.0223620-turtle.txt
N-Triples: 52383-1.0223620-rdf-ntriples.txt
Original Record: 52383-1.0223620-source.json
Full Text

Full Text

Quantification of lung surface area usingcomputed tomographyYuan et al.Yuan et al. Respiratory Research 2010, 11:153http://respiratory-research.com/content/11/1/153 (31 October 2010)RESEARCH Open AccessQuantification of lung surface area usingcomputed tomographyRen Yuan1,2, Taishi Nagao1, Peter D Paré1,3, James C Hogg1,4, Don D Sin1, Mark W Elliott1, Leanna Loy1, Li Xing1,Steven E Kalloger1, John C English5, John R Mayo2, Harvey O Coxson1,2*AbstractObjective: To refine the CT prediction of emphysema by comparing histology and CT for specific regions of lung.To incorporate both regional lung density measured by CT and cluster analysis of low attenuation areas forcomparison with histological measurement of surface area per unit lung volume.Methods: The histological surface area per unit lung volume was estimated for 140 samples taken from resectedlung specimens of fourteen subjects. The region of the lung sampled for histology was located on the pre-operative CT scan; the regional CT median lung density and emphysematous lesion size were calculated using theX-ray attenuation values and a low attenuation cluster analysis. Linear mixed models were used to examine therelationships between histological surface area per unit lung volume and CT measures.Results: The median CT lung density, low attenuation cluster analysis, and the combination of both wereimportant predictors of surface area per unit lung volume measured by histology (p < 0.0001). Akaike’s informationcriterion showed the model incorporating both parameters provided the most accurate prediction of emphysema.Conclusion: Combining CT measures of lung density and emphysematous lesion size provides a more accurateestimate of lung surface area per unit lung volume than either measure alone.BackgroundThe major pathological components responsible for thedecrease in maximal expiratory flow that characterizeChronic Obstructive Pulmonary Disease (COPD) includean increase in airway resistance due to small airwayremodeling and obliteration, and a decrease in elasticrecoil secondary to the parenchymal tissue destructionwhich characterizes emphysema [1-3]. Separating thecontribution of each of these two components can pro-vide better understanding of the natural history of dis-ease, allow monitoring of disease progression, evaluatethe impact of a therapeutic intervention and potentiallyguide the most appropriate therapeutic target in indivi-dual patients. The fact that pulmonary function testscannot separate these structural changes [4], andbecause pathological estimates can only do so in surgicalor postmortem specimens, has led to attempts to usechest CT scans to measure these changes in vivo.A number of quantitative CT lung densitometry mea-surements have been employed to measure the extent ofemphysema including, 1) the relative area of lung withattenuation values lower than various thresholds [5-10],2) a specific percentile point on the frequency-attenua-tion distribution curve [8,9,11], and 3) median lunginflation [12]. However, measurement of lung densitymay not be the most efficient way to detect emphysemaif tissue destruction is accompanied by “remodeling” ofthe lung parenchyma, such as fibrosis [13-15]. Mishimawas the first to introduce cluster analysis of low attenua-tion areas - a method to measure the size distribution oflow attenuation regions [16]. Although validation of thisparameter against pathologic standards is controversial[8], we postulated that cluster analysis would supple-ment lung densitometry in the detection and quantifica-tion of emphysema since it is less likely to be affectedby tissue deposition.In the present study, we tested the relationship betweenthe histopathologic reference standard for emphysema -airspace surface area per unit lung volume (SA/V), andtwo CT measurements: CT lung densitometry (median* Correspondence: Harvey.Coxson@vch.ca1University of British Columbia James Hogg Research Centre and the Heartand Lung Institute, St. Paul’s Hospital; Burrard Street, Vancouver, CanadaFull list of author information is available at the end of the articleYuan et al. Respiratory Research 2010, 11:153http://respiratory-research.com/content/11/1/153© 2010 Yuan et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative CommonsAttribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction inany medium, provided the original work is properly cited.lung density) and CT cluster analysis. We hypothesizedthat the combination of the two CT measurements willbe superior to the sole use of either in the prediction ofSA/V.MethodsSubject SelectionFourteen subjects (9 men, 5 women) were included inthe present study (Table 1). Ten patients underwentlobectomy and four underwent pneumonectomy forlung cancers. Preoperatively, all subjects had spirometrymeasurements and the diffusing capacity (DLco) wasmeasured by the single-breath method of Miller andassociates [17]. The study was approved by the hospitaland university ethical review boards and all subjectsprovided written informed consent for the use of allmaterials and data.CT TechniqueAll subjects received a pre-operative, non-contrast heli-cal CT scan in the supine position at the end of fullinspiration. 11 subjects were scanned using a GE Light-Speed Ultra CT scanner (General Electric Medical Sys-tems, Milwaukee, WI) with the following settings: 120kVp, 114 mAs, and 5 mm slices thickness; and 3 sub-jects were scanned using a Siemens Sensation 16 CTscanner (Siemens Medical Solutions; Erlangen, Ger-many) with the following parameters: 120 kVp, 115mAs, and 5 mm slice thickness. The scanners were cali-brated regularly using standard water and air phantomsto allow for comparisons between individuals andbetween scanners.Quantitative HistologyFollowing surgery, the resected specimen was trans-ferred directly from the operating room to the labora-tory. The specimen was inflated with Bouin fixative at aconstant distending pressure of 25 cm of water andimmersed in formalin overnight. After fixation, eachspecimen was cut into ten slices with 5-8 mm thicknessin the axial plane and photographed using a digital cam-era (Nikon Coolpix, Nikon Corp., Japan). A grid of 2 ×2 cm squares was superimposed over each lung slice,one square was randomly selected and the tissuebeneath it was excised, embedded in paraffin, sectionedand stained with haematoxylin and eosin, which resultedin 140 tissue samples in total. Ten random images perhistology section were captured using a light microscope(Nikon Microphot) equipped with a digital camera(JVC3-CCD KY F-70, Diagnostic Instruments). The digi-tal images were analyzed using stereologic techniquesand a custom program written for Image Pro Plus® digi-tal-image-analysis software (Media Cybernetics) asdescribed elsewhere [18]. Briefly, each image was binar-ized and a grid of lines was superimposed on the image.The program automatically counts the number of inter-sections between the superimposed lines and the alveo-lar walls (i.e., tissue-air interface), the number of lineendpoints in one image (i.e., ΣP total), as well as thenumber of line endpoints that fall on tissue (i.e., ΣPtis-sue). Surface area per unit lung volume (SA/V) was cal-culated using the following equations as previouslydescribed [12]:( / )SA V surface density of the tissue air interfacevolume =×−fraction of tissue,(1)in which, surface density of the tissue-air interface[19]:Sv tis 4 L I Ptissue 2 mean linear intercept( ) = ( ) × ( ) =/ / /Σ Σ (2)where L = the length of the grid unit line, ΣI = thenumber of intersections counted, ΣP tissue is the num-ber of line end points that fall on tissue.The volume fraction of tissue:Vv tis P tissue P total( ) = Σ Σ/ , (3)where ΣP total is the number of line end pointscounted in one image.SA/V for each of the samples was corrected forshrinkage. The shrinkage factor was determined by mea-suring the length of one side of the blocks prior to fixa-tion processing and then dividing by the length of thatside of the cut sections post-fixation (shrinkage factor:1.30 ± 0.13)Quantitative CTThe region of lung where the histology samples weretaken was identified on the CT image by comparinganatomic landmarks on the cut surface of the gross lungspecimen and CT images as shown in Figure 1. TheTable 1 Subjects DemographicsMean ± SD RangeAge (yrs) 67.0 ± 3.1 61.8 - 72.0Gender 5 female:9 maleSmoking (pack yrs) 59.6 ± 44.4 24.8 - 173.0Height (cm) 169.1 ± 7.2 157.0 - 180.0Weight (kg) 66.6 ± 12.5 44.0 - 90.0Post-FEV1%pred (%) 78.7 ± 16.1 46.7 - 114.5Post-FEV1/FVC 67.5 ± 8.8 45.9 - 79.0DLCO % pred 70.4 ± 10.3 47.8 - 90.6Post-FEV1%pred: post-bronchodilator forced expiratory flow in one second/predicted value.Post-FEV1/FVC: post-bronchodilator forced expiratory flow in one second/post-bronchodilator forced vital capacity.DLco: Diffusing capacity.Yuan et al. Respiratory Research 2010, 11:153http://respiratory-research.com/content/11/1/153Page 3 of 9difference in lung inflation between the in vivo and invitro state was determined by comparing the area of thecut surface on the lung specimen, measured using Ima-geJ, (Rasband, W.S., ImageJ, U. S. National Institutes ofHealth, Bethesda, Maryland, USA, http://rsb.info.nih.gov/ij/, 1997-2007) to the area of the lung on thein vivo CT image measured using custom software(EmphylxJ, UBC James Hogg Research Centre, Vancou-ver, B.C, http://www.flintbox.com) as described else-where [20]. Then, a square, size-corrected for inflationwas superimposed upon the CT image. For each voxelwithin that square, the apparent X-ray attenuation value(Hounsfield Unit, HU) was obtained and converted togravimetric density (g/ml) by adding 1000 to the HUvalue and dividing by 1000 [21]. The median CT lungdensity value was chosen from the frequency distribu-tion curve of lung density within each square since thecurve is skewed to the right [12]. We estimated the dis-tribution of sizes of the emphysematous lesions withineach square using a low attenuation cluster analysis[16,22]. In the low attenuation cluster analysis theinverse slope of the log-log relationship of the size ofthe low attenuation cluster (number of contiguous vox-els <-856 HU) versus the number of clusters of that sizeis the power-law exponent (D). -856HU was chosen toidentify “emphysematous” because it converts to 6.0 ml/g, which has been previously shown to represent theboundary between normal and mildly emphysematouslung [12] (See additional file 1: Converting 6.0 ml/g to-856HU).Statistical AnalysisThe primary outcome was the histologically measuredSA/V and the independent variables included the med-ian CT lung density and the CT cluster analysis valueD. We used a linear mixed model (the REstricted Maxi-mum Likelihood method, REML) to incorporate thewithin subject variance of the measurements since tenmeasurements were made from each lung specimen[23], and we examined the association between the out-come and the two independent variables with the gen-der, age and patient’s body mass index (BMI) beingcovariates. To test whether CT cluster analysis couldsupplement lung densitometry (i.e., median lung density)in detecting histological emphysema, we compared theprediction of SA/V using median CT lung density orthe CT cluster analysis value D to a third model, whichincorporated both variables using Akaike’s InformationCriterion (AIC) based on the Maximum Likelihood Esti-mation [24]. The model with the smallest AIC value isconsidered to be the best model [25]. Analyses wereperformed using SAS version 9.1 (Carey, N.C.). Statisti-cal significance was defined at a p-value less than 0.05.Continuous variables are expressed as mean ± SD.ResultsSubject CharacteristicsThe subject demographics are shown in Table 1. Thelevel of airway obstruction of the subjects was relativelymild with only one subject in stage 3 according to theGlobal Initiative for Obstructive Lung Disease (GOLD)categories [26]. Five subjects were stage 2, two stage 1,and the remaining six subjects had normal lungfunction.Quantitative Histology and Quantitative CTMeasurementsThe histological measurements of SA/V and quantitativeCT measurements for all 140 tissue samples from 14cases are summarized in Table 2. These data show thatthere is a wide variation in both histological and quanti-tative CT measurements within each individual.Linear mixed models showed that the median CT lungdensity and the CT cluster analysis value D were signifi-cantly associated with histological SA/V (both p <0.0001) (Figures 2 and 3). The prediction equations ofSA/V using CT lung density alone, CT cluster analysisalone, and the combination of these two measurementswere:SA/V = 4.62 + 1631.99 × median CT lung density;SA/V = 168.44 + 69.21 × CT cluster analysis value D;Figure 1 Matching CT Images and Lung Specimens. A CT imageof a representative subject is shown in Figure 1A and thecorresponding slice of the resected specimen is shown in Figure 1B.For reference and orientation, the tumor is marked by a star (*). Agrid is superimposed over the fixed lung slice (Figure 1B) and a 2 ×2 cm square section (square E) is randomly selected for histologicalprocessing and measurement of surface area per unit lung volume(SA/V). The corresponding region (square E) on CT is then identified(Figure 1A); the CT median lung density and the CT cluster analysisvalue D are obtained in the region of interest using the computerprogram (EmphylxJ). The size of the square E on CT has beencorrected for lung inflation to match the size of the histologicalspecimen.Yuan et al. Respiratory Research 2010, 11:153http://respiratory-research.com/content/11/1/153Page 4 of 9SA/V = 6.04 + 1597.05 × median CT lung density +11.19 × CT cluster analysis value D.A comparison of the three models using the Akaike’sInformation Criterion showed that the model incorpor-ating both CT lung density and low attenuation clusteranalysis yielded the smallest AIC value indicating thatit is the best model for predicting SA/V (the AIC was904 for CT lung density alone, 927 for CT cluster ana-lysis alone and 897 for the model incorporating bothvariables).DiscussionThe most important finding of the present study is thatalthough CT lung densitometry (i.e., median lung den-sity in the current study) was a valid estimate of the his-tological measurement of airspace enlargement and/oralveolar wall destruction (airspace surface area per unitlung volume, SA/V), its accuracy was significantlyimproved by combining it with CT cluster analysis oflower attenuation areas. Basing an estimate of emphy-sema solely on a measure of lung density assumes thatthe decrease in alveolar surface area which accompaniesemphysema is mirrored by a proportional reduction inlung tissue mass. Although it is clear that tissue destruc-tion is part of the process, there is increasing evidencethat emphysema is also accompanied by “remodeling” ofthe lung parenchyma which may be associated withfibrosis [13-15]. The extent of this “remodeling” willconfound the relationship between lung density and SA/V. This phenomenon is illustrated in Figure 4. In thisschematic, normal lung architecture (Normal) and twoexamples of “emphysema” (A and B) are shown. Inexample A, there is a loss of alveolar walls with a corre-sponding loss of lung mass. In example B, there is asimilar loss of the number of alveolar walls but a thick-ening of the retained alveolar walls such that the massof the lung is comparable to Normal and greater than inA although both A and B have comparable loss in lungSA/V.CT cluster analysis of low attenuation areas is amethod to describe and quantify the distribution ofemphysematous spaces by determining whether lowTable 2 Histological and Quantitative CT Measurementsfor 140 Tissue Samples from 14 SubjectsSubject Histology-SA/V(cm2/cm3)Median CTlung density(g/ml)Low AttenuationClusterAnalysis (D)1 161.4 ~ 275.3 5.6 ~ 7.9 0.2 ~ 1.12 175.1 ~ 265.6 6.5 ~ 7.5 0.1 ~ 0.73 102.5 ~ 215.3 5.9 ~ 8.3 0.2 ~ 0.94 182.7 ~ 438.6 4.2 ~ 5.8 0.6 ~ 2.55 39.2 ~ 122.2 11.7 ~ 39.1 0.1 ~ 0.36 172.0 ~ 253.9 4.7 ~ 6.9 0.2 ~ 1.27 84.3 ~ 171.3 8.2 ~ 14.8 0.1 ~ 0.48 171.9 ~ 289.2 5.6 ~ 9.3 0.3 ~ 1.29 90.6 ~ 260.1 7.3 ~ 13.8 0.1 ~ 0.610 227.4 ~ 464.1 2.9 ~ 4.8 1.1 ~ 2.011 141.7 ~ 256.5 3.2 ~ 6.7 0.6 ~ 2.012 320.2 ~ 445.6 3.6 ~ 5.9 0.9 ~ 2.213 78.0 ~ 248.3 6.1 ~ 14.8 0.1 ~ 0.714 237.6 ~ 332.6 4.8 ~ 6.3 0.6 ~ 2.0Figure 2 Association between the Histological SA/V and CTMedian Lung Density. There is a significant association betweenthe SA/V (cm2/cm3) measured histologically and the CT medianlung density (g/ml) (r = 0.82, p < 0.0001). All subjects are shownusing different symbols. Data point A and B refer to samples withcomparable SA/V value but very different CT density measurement(sample A: SA/V = 247 cm2/cm3, CT density = 0.14 g/ml; sample B:SA/V = 258 cm2/cm3, CT density = 0.24 g/ml). A and B refer to thesame samples in Figure 2, 3, and 5.Figure 3 Association between the Histological SA/V and CTCluster Analysis Value D. There is a significant associationbetween the SA/V (cm2/cm3) measured histologically and the CTcluster analysis D value (r = 0.74, p < 0.0001). All subjects are shownusing different symbols. Data point A and B have comparable valuefor SA/V and CT cluster analysis (sample A: SA/V = 247 cm2/cm3, D= 0.91; sample B: SA/V = 258 cm2/cm3, D = 1.17). A and B refer tothe same samples in Figure 2, 3, and 5.Yuan et al. Respiratory Research 2010, 11:153http://respiratory-research.com/content/11/1/153Page 5 of 9attenuation voxels are clustered into large lesions orpresent as diffuse small ones. It has been shown thatthere is an inverse power law relationship between thesize and number of clusters where the slope of this rela-tionship (D) becomes smaller with increasing lesion size[16]. This variable is less likely to be affected by theaccumulation of connective tissue that may accompanyemphysema since it measures clustering of low attenua-tion areas. Examples of these theoretical considerationswere observed in our data. For example, points A and Bin Figure 2 and 3 represent two samples with compar-able values for histological SA/V and CT cluster analysisbut very different CT lung density. The examination ofthe histology in these two samples shown in Figure 5 isconsistent with the theory illustrated in Figure 4. Forsample B CT cluster analysis provides a more accurateestimate of histological SA/V than does CT lung density,because tissue deposition accompanies tissue destruc-tion. Additionally the cluster analysis likely detects truetissue destruction with the formation of low attenuationareas larger than single CT voxels while measures ofdensity can be affected by simple hyperinflation of lungtissue without alveolar wall destruction. Such hyperinfla-tion may be a precursor of the tissue destruction whichcharacterizes emphysema but would have less effect onthe histological surface area to volume ratio than truetissue disruption.The current data also suggest that the cluster analysisvalue D, per se, is a valid quantitative CT estimate ofemphysema because it significantly, and independently,correlated with the histological measurement of surfacearea per unit lung volume (Figure 3). This finding is atvariance with that of Madani et al [8]. We think thisdiscrepancy might be because we chose a different HUcutoff to define the “low attenuation cluster”. Madaniet al chose -960HU and 1st percentile point as the cutoffwhereas we used a relatively higher HU value: -856HU.As we explained in the methods section that -856 HU isconverted from a lung tissue inflation value of 6.0 ml/g,which was previously shown to represent the boundarybetween normal and mild emphysematous lung [12].In the current study, we chose surface area per unitlung volume (i.e., SA/V) as the histological reference.This variable has been shown to separate normal lungfrom emphysematous tissue [12], and its calculation(Equation 1 and 2) is linearly related to the mean linearintercept (i.e., Lm), which has been used by other groupsto estimate emphysema microscopically [9].One challenge for validation of CT measurements isthe marked heterogeneity of the emphysematous process.Even in lungs severely affected by emphysema, someregions still maintain normal architecture making sam-pling for pathological examination critical as shown inFigure 6. In many of the previous validation studies,Figure 4 A Schematic Showing the Relationship between LungSA/V and Density under two scenarios. The top panel representsnormal lung architecture with the dimensions of each “alveolus” being100 × 100 μm yielding a total volume of the “lung” = 16,000 μm3 witha surface area of 6,400 μm2 and a SA/V of 0.4. If we assign a mass of 10units to each 100 μm length of “alveolar wall” this “lung” has a mass of400 units and a density of 0.025 units/μm3 (= 400 units/16,000 μm3). InA, the volume and thickness of the “alveolar walls” remains the sameas those in “normal lung architecture” but the surface area is decreaseddue to destruction of “alveolar walls”. In this scenario, the reduction inSA/V and density are proportional. However in scenario B, thethickness of the “alveolar walls” is doubled therefore increasing themass. The resultant SA/V is the same as in A whereas the density ishigher than in A and even higher than the Normal. Thus if there isaddition of tissue, the relationship between SA/V and density isdisrupted.Figure 5 Hematoxylin and Eosin-stained Images of TissueSamples A and B in Figures 2, 3. The tissue shown in A has a SA/V of 247 mm2/mm3 and a CT density of 0.14 g/ml while the area inB has a SA/V of 258 mm2/mm3 and a CT density of 0.24 g/ml. Thusdespite comparable SA/V, there is a substantial difference in CTdensity due to the deposition of extracellular matrix in B. On theother hand, CT cluster analysis (i.e., value D), which relies solely onthe size of the low attenuation areas, was comparable in these tworegions (0.97 in A and 1.17 in B).Yuan et al. Respiratory Research 2010, 11:153http://respiratory-research.com/content/11/1/153Page 6 of 9including our previous work, the commonly appliedapproach is to randomly sample tissue from lungs, calcu-late the averaged value from these random samples toobtain one single histological measurement for each sub-ject, and compare this value to one single CT measure-ment obtained from the whole lung of that subject[6,8,9,11,12]. However, by doing so, the CT measurementis global, incorporating all regions, diseased or relativelynormal, whereas the histological measurement is aver-aged from a limited number of samples taken from differ-ent regions of the surgically resected lungs. In the presentstudy, we have refined this approach by using a modifiedcomputer program, which enables us to obtain regionalCT measurements from the exact regions of the lungwhere the histological measurements were taken andcompare this regional CT measurement to the histologi-cal measurement of the same region. We think this pre-cise matching can provide a more accurate comparisonbetween CT and histological measurements. Also, in thisway, we were testing our hypothesis in 140 tissue samplesrather than in 14 subjects. Nevertheless, we cannot con-sider 140 tissue samples as 140 independent samplessince ten samples were taken from each individual.Therefore, in the statistical analysis, we applied a linearmixed modeling approach to account for the randomeffects arising from inter-individual variance and toobtain prediction equations at the group level [23].This study has some limitations. First, in the currentstudy, we only used one CT densitometry measurement,median lung density. While Gevenois has shown usingthin slice CT scans (1 mm) that -950 HU detects bothmacroscopic and microscopic emphysema they alsoshowed that using this cut-off 6.8% would be the upperlimit of normal and therefore the threshold betweennormal and diseased [6]. However, previous studiesusing thick slice CT scans shows that threshold cut-offssuch as -910 HU only pick up large emphysematousholes in the lung [27] while a threshold of -856 HUestimates the small holes [12]. Therefore, with this datain mind, we chose the mean lung density threshold,because of the small size of pathologic specimens (2 × 2cm2) that we were comparing to the thick slice CTvalues and the relatively mild degree of emphysema pre-sent in our subjects and specimens. We cannot com-ment on the supplementary role of CT cluster analysisto other more traditional whole lung CT densitometrymeasurements of emphysema, such as low attenuationarea and percentile point, etc. However, we believe it isreasonable to assume that CT cluster analysis wouldsupplement the other CT densitometry measurementssince all such measurements rely on choosing a cutoffvalue from the X-ray attenuation distribution histogram,either along the X axis (i.e., low attenuation area) oralong the Y axis (i.e., percentile point). The extent, towhich, CT cluster analysis supplements the different CTdensitometry measurements might vary depending onthe threshold use and, therefore, further studies includ-ing other densitometry measurements may providemore information. Secondly, we used -856HU, based onour previous experience with thick slice CT scans thatidentified this HU threshold as effective in identifyingmild emphysematous areas [12]. We realize that CTscan slices in our previous study were of 10 mm thick-ness whereas in the current study were of 5 mm slicethickness. Due to limitations in CT scanner technology,we are not able to test whether this threshold is equallyeffective using either slice thickness. Lastly, the pre-surgery CT images were acquired using two differentCT scanners could have introduced errors in CT lungdensity measurement. However since the X-ray radiationdose is similar (120 kVp and 114 mAs on GE scanner;120 kVp and 115 mAs on Siemens scanner), we believethis effect is small. Moreover we have previously shownthat CT densitometry measurements using similaracquisition protocols are comparable between these CTscanners [20].The difference in Akaike’s Information Criterion (AIC)between the models appears small but this does notmean that the added information of the combinedmodel is small. The AIC cannot be interpreted using atraditional “hypothesis testing” statistical paradigm. Itdoes not generate a P value, does not reach conclusionsabout “statistical significance”, and does not “reject” anymodel. AIC determines how well the data supports eachmodel, taking into account both the goodness-of-fit(sum-of-squares) and the number of parameters in themodel. Ultimately, the model with the smallest AIC isconsidered the best, although the AIC value itself is notmeaningful [28].In conclusion, the results of this study show that anaccurate comparison between CT and histological mea-surements can be achieved by precisely mapping theFigure 6 Heterogeneity of Lung Tissue Destruction. Examples ofhematoxylin and eosin-stained images of tissue samples taken fromthe same individual but different lung regions. A: Normal tissue withSA/V = 439 cm2/cm3, tissue density = 0.19 g/ml, B: emphysematoustissue with SA/V = 183 cm2/cm3, tissue density = 0.14 g/ml.Yuan et al. Respiratory Research 2010, 11:153http://respiratory-research.com/content/11/1/153Page 7 of 9location of the histological sample to its in vivo locationon the CT. In addition, the CT cluster analysis value Dcan supplement CT densitometry in detecting and quan-tifying emphysema. The additional benefit may be dueto the fact that cluster analysis is more sensitive to truetissue destruction and immune to the artifact caused bythe deposition of connective tissue that may accompanythe emphysematous process.Additional materialAdditional file 1: Conversion of 6.0 ml/g to -856HU. This file outlinesthe method to convert lung inflation values, measured as ml of gas perg tissue, into X-ray attenuation values.AcknowledgementsThe authors thank Anh-Toan Tran, BSc and Ida Chan, MD for technicalassistance in developing and supporting the lung analysis application.PDP is a Michael Smith Foundation for Health Research Distinguishedscholar and the Jacob Churg Distinguished Researcher. DDS is a CanadaResearch Chair in COPD and a Senior Scholar with the Michael SmithFoundation for Health Research. HOC was Parker B Francis Fellow inPulmonary Research during the time of this research. HOC is currently aCanadian Institutes of Health Research (CIHR)/British Columbia LungAssociation New Investigator. HOC is also supported, in part, by theUniversity of Pittsburgh COPD SCCOR NIH 1P50 HL084948 and R01HL085096 from the National Heart, Lung, and Blood Institute, NationalInstitutes of Health, Bethesda, MD to the University of Pittsburgh. Thisproject was funded by a CIHR Industry partnership grant withGlaxoSmithKline.Author details1University of British Columbia James Hogg Research Centre and the Heartand Lung Institute, St. Paul’s Hospital; Burrard Street, Vancouver, Canada.2UBC Department of Radiology, Vancouver General Hospital; West 12th Ave.Vancouver, Canada. 3UBC Department of Medicine St. Paul’s Hospital; BurrardStreet, Vancouver, Canada. 4UBC Department of Pathology, St. Paul’sHospital; Burrard Street, Vancouver, Canada. 5UBC Department of Pathology,Vancouver General Hospital, West 12th Ave. Vancouver, Canada.Authors’ contributionsRY and TN carried out the quantitative CT analysis. WME and LL carried outthe quantitative histological analysis. DS and LX performed the statisticalanalysis. PP is the principal investigator of the project, obtained funding forand supervised the project. PP, JH, and HC participated in the design of thestudy. RY, PP, JH and HC drafted the manuscript. SK, JE and JM participatedin the coordination of the study and helped to draft the manuscript. Allauthors read and approved the final manuscript.Competing interestsPD Paré is the principal investigator of a project funded by GSK to developCT based algorithms to quantify emphysema and airway disease in COPD.With collaborators he has received ~ $300,000 to develop and validate thesetechniques. These funds he have been applied solely to the research tosupport programmers and technicians. Peter Pare was also PI of a MerckFrosst supported research program to investigate gene expression in thelungs of patients who have COPD. He and collaborators have received ~$200,000 for this project. These funds have supported the technicalpersonnel and expendables involved in the project. PP has established anew contract with Merck to discover genetic predictors of gene expressionin lung tissue. With collaborators he will receive $95,000 over the next yearto do this work. The funds will support personnel and buy supplies. PP sitson an advisory board for Talecris Biotherapeutics who make anti-oneantitrypsin replacement therapy.JC Hogg has served as a consultant, given lectures and participated inadvisory boards of several major pharmaceutical companies in the past fiveyears. The total reimbursement for these activities is less than $20000.00. HisUniversity (UBC) has also received industry sponsored grants from GSK andMerck on which he serve as the PI.DD Sin has received research funding from GlaxoSmithKline andAstraZeneca for projects on chronic obstruction pulmonary disease. DD Sinhas also received honoraria for speaking engagements for talks on COPDsponsored by these organizations.HO Coxson received $4800 in 2006 - 2008 for serving on the steeringcommittee for the ECLIPSE project for GSK. In addition HC is the co-investigator on two multi-center studies sponsored by GSK and has receivedtravel expenses to attend meetings related to the project. HC has threecontract service agreements with GSK to quantify the CT scans in subjectswith COPD and a service agreement with Spiration Inc to measure changesin lung volume in subjects with severe emphysema. A percentage of HC’ssalary between 2003 and 2006 (15,000 US $/year) derives from contractfunds provided to a colleague PD Pare by GSK for the development ofvalidated methods to measure emphysema and airway disease usingcomputed tomography. HC is the co-investigator (DD Sin PI) on a CanadianInstitutes of Health - Industry (Wyeth) partnership grant.R Yuan, T Nagao, WM Elliott, L Loy, L Xing, S Kalloger, J English, and J Mayohave no competing interests in the content of this manuscript.Received: 15 June 2010 Accepted: 31 October 2010Published: 31 October 2010References1. Burrows B, Knudson RJ, Cline MG, Lebowitz MD: Quantitative relationshipsbetween cigarette smoking and ventilatory function. Am Rev Respir Dis1977, 115:195-205.2. Hogg JC, Macklem PT, Thurlbeck WM: Site and nature of airwayobstruction in chronic obstructive lung disease. N Engl J Med 1968,278:1355-1360.3. Macklem PT, Mead J: Resistance of central and peripheral airwaysmeasured by a retrograde catheter. J Appl Physiol 1967, 22:395-401.4. Fraser RS, Paré PD, Colman NC, Muller NL: Diagnosis of Diseases of the ChestPhiladelphia: Saunders, Fourth 1999.5. Bankier AA, De Maertelaer V, Keyzer C, Gevenois PA: Pulmonaryemphysema: subjective visual grading versus objective quantificationwith macroscopic morphometry and thin-section CT densitometry.Radiology 1999, 211:851-858.6. Gevenois PA, De Vuyst P, de Maertelaer V, Zanen J, Jacobovitz D, Cosio MG,Yernault JC: Comparison of computed density and microscopicmorphometry in pulmonary emphysema. Am J Respir Crit Care Med 1996,154:187-192.7. Hayhurst MD, MacNee W, Flenley DC, Wright D, McLean A, Lamb D,Wightman AJ, Best J: Diagnosis of pulmonary emphysema bycomputerised tomography. Lancet 1984, 2:320-322.8. Madani A, Van Muylem A, de Maertelaer V, Zanen J, Gevenois PA:Pulmonary emphysema: size distribution of emphysematous spaces onmultidetector CT images-comparison with macroscopic and microscopicmorphometry. Radiology 2008, 248:1036-1041.9. Madani A, Zanen J, de Maertelaer V, Gevenois PA: Pulmonary emphysema:objective quantification at multi-detector row CT–comparison withmacroscopic and microscopic morphometry. Radiology 2006,238:1036-1043.10. Muller NL, Staples CA, Miller RR, Abboud RT: “Density mask”. An objectivemethod to quantitate emphysema using computed tomography. Chest1988, 94:782-787.11. Gould GA, MacNee W, McLean A, Warren PM, Redpath A, Best JJ, Lamb D,Flenley DC: CT measurements of lung density in life can quantitate distalairspace enlargement - an essential defining feature of humanemphysema. Am Rev Respir Dis 1988, 137:380-392.12. Coxson HO, Rogers RM, Whittall KP, D’Yachkova Y, Pare PD, Sciurba FC,Hogg JC: A quantification of the lung surface area in emphysema usingcomputed tomography. Am J Respir Crit Care Med 1999, 159:851-856.13. Lang MR, Fiaux GW, Gillooly M, Stewart JA, Hulmes DJ, Lamb D: Collagencontent of alveolar wall tissue in emphysematous and non-emphysematous lungs. Thorax 1994, 49:319-326.Yuan et al. Respiratory Research 2010, 11:153http://respiratory-research.com/content/11/1/153Page 8 of 914. Tonelli M, Stern EJ, Glenny RW: HRCT evident fibrosis in isolatedpulmonary emphysema. J Comput Assist Tomogr 1997, 21:322-323.15. Cardoso WV, Sekhon HS, Hyde DM, Thurlbeck WM: Collagen and elastin inhuman pulmonary emphysema. Am Rev Respir Dis 1993, 147:975-981.16. Mishima M, Hirai T, Itoh H, Nakano Y, Sakai H, Muro S, Nishimura K, Oku Y,Chin K, Ohi M, et al: Complexity of terminal airspace geometry assessedby lung computed tomography in normal subjects and patients withchronic obstructive pulmonary disease. Proc Natl Acad Sci USA 1999,96:8829-8834.17. Miller A, Thornton JC, Warshaw R, Anderson H, Teirstein AS, Selikoff IJ:Single breath diffusing capacity in a representative sample of thepopulation of Michigan, a large industrial state. Predicted values, lowerlimits of normal, and frequencies of abnormality by smoking history. AmRev Respir Dis 1983, 127:270-277.18. Hogg JC, Chu F, Utokaparch S, Woods R, Elliott WM, Buzatu L,Cherniack RM, Rogers RM, Sciurba FC, Coxson HO, Pare PD: The nature ofsmall-airway obstruction in chronic obstructive pulmonary disease. NEngl J Med 2004, 350:2645-2653.19. Howard CV, Reed MG: Unbiased Stereology: Three-Dimensional Measurementin Microscopy, Second Edition Summary Liverpool, UK: Taylor & Francis Inc,Second 2004.20. Yuan R, Mayo JR, Hogg JC, Pare PD, McWilliams AM, Lam S, Coxson HO:The Effects of Radiation Dose and CT Manufacturer on Measurements ofLung Densitometry. Chest 2007, 132:617-623.21. Hedlund LW, Vock P, Effmann EL: Evaluating lung density by computedtomography. Semin Respir Med 1983, 5:76-87.22. Coxson HO, Whittall KP, Nakano Y, Rogers RM, Sciurba FC, Keenan RJ,Hogg JC: Selection of patients for lung volume reduction surgery usinga power law analysis of the computed tomographic scan. Thorax 2003,58:510-514.23. Feldman HA: Families of lines: random effects in linear regressionanalysis. J Appl Physiol 1988, 64:1721-1732.24. Verbeke G, Molenberghs G: Linear Mixed Models for Longitudinal DataSpringer-Verlag New York 2000.25. Ljung L: System Identification: Theory for the User Upper Saddle River, NJ:Prentice-Hal PTR 1999.26. Rabe KF, Hurd S, Anzueto A, Barnes PJ, Buist SA, Calverley P, Fukuchi Y,Jenkins C, Rodriguez-Roisin R, van Weel C, Zielinski J: Global strategy forthe diagnosis, management, and prevention of chronic obstructivepulmonary disease: GOLD executive summary. Am J Respir Crit Care Med2007, 176:532-555.27. Miller RR, Muller NL, Vedal S, Morrison NJ, Staples CA: Limitations ofcomputed tomography in the assessment of emphysema. Am Rev RespirDis 1989, 139:980-983.28. Lindsey JK, Jones B: Choosing among generalized linear models appliedto medical data. Stat Med 1998, 17:59-68.doi:10.1186/1465-9921-11-153Cite this article as: Yuan et al.: Quantification of lung surface area usingcomputed tomography. Respiratory Research 2010 11:153.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitYuan et al. Respiratory Research 2010, 11:153http://respiratory-research.com/content/11/1/153Page 9 of 9


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items