The Prediction of Small Airway Dimensions UsingComputed TomographyYasutakaNakano, JonathanC. Wong, PimA. deJong, LillianaBuzatu, Taishi Nagao, Harvey O. Coxson,W. Mark Elliott, James C. Hogg, andPeter D. PareDepartment of Respiratory Medicine, Shiga University of Medical Science, Otsu, Shiga, Japan; James Hogg iCAPTURE Centre for Cardiovascularand Pulmonary Research, St. Pauls Hospital; and Department of Radiology, Vancouver General Hospital, University of British Columbia,Vancouver, British Columbia, CanadaChronic obstructive pulmonary disease is characterized by destruc-tionofthelungparenchymaand/orsmall airwaynarrowing. Todeterminewhether thedimensionsofrelativelylargeairwaysas-sessedusingcomputedtomography(CT)reflectsmall airwaydi-mensionsmeasuredhistologically,weassessedthesevariablesinnonobstructedormildtomoderatelyobstructedpatientshavinglobar resection for a peripheral tumor. For both CT and histology,the square root of the airway wall area (Aaw) was plotted versuslumen perimeter to estimate wall thickness. The wall area percent-agewascalculatedaswall area/lumenareawall area100.AlthoughCToverestimatedAaw, theslopesoftherelationshipsbetween the square root of Aaw and internal perimeter (Pi) mea-sured with both techniques were related (CT slope 0.2059 histol-ogyslope0.1701, R20.32, p0.01). Themeanwall areapercentage measured by CT for airways with a Pi of greater than0.75 cm predicted the mean dimensions of the small airways withan internal diameter of 1.27 mm (R20.57, p 0.01). We concludethatCTmeasurementsofairwayswithaPi of0.75cmormorecould be used to estimate the dimensions of the small conductingairways, which are the site of airway obstruction in chronic obstruc-tive pulmonary disease.Keywords: bronchioles; bronchiolitis; chronicobstructivepulmonarydisease; emphysema; small airwaysChronic obstructive pulmonary disease (COPD) is characterizedbydecreasedmaximal expiratoryairow, hyperination, andgastrapping.Thesephysiologicabnormalitiesarecausedbyacombination of loss of lung elastic recoil and narrowing of thesmall airways. Emphysema is the pathologic lesion most closelyassociatedwiththelossof elasticityandincreasedtotal lungvolume, whereas inammation and brosis of the membranousbronchioles accompanied by mucous plugging characterize thepathologic lesions that contribute to the small airway narrowing(14). Together these abnormalities cause increased airway resis-tance and premature airway closure. Several studies have estab-lished (2, 5, 6) that the major site of airway obstruction in patientswith COPD is in airways smaller than 2-mm internal diameter,and a recent histologic evaluation of a large group of cases has(Received in original form July 6, 2004; accepted in final form October 25, 2004)Supportedby fundingfrom theNational Institutesof Health(NHLBI HL64068-03), GlaxoSmithKline, and a Gerrit Jan Mulder Stichting Fellowship (P.A.d.J.) (GJM-stichting, Erasmus Medical Centre, Rotterdam, TheNetherlands) (H.O.C. is aParker B. Francis Fellow).Correspondence and requests for reprints should be addressed to Peter D. Pare,M.D., James Hogg iCAPTURE Centre for Cardiovascular and Pulmonary Research,University of British Columbia, St. Pauls Hospital, 1081 Burrard Street, Room166,Vancouver, BC, V6Z 1Y6 Canada. E-mail: ppare@mrl.ubc.caThis article has an online supplement, which is accessible from this issues tableof contents at www.atsjournals.orgAm J Respir Crit Care Med Vol 171. pp 142146, 2005Originally Published in Press as DOI: 10.1164/rccm.200407-874OC on October 29, 2004Internet address: www.atsjournals.orgshown that the decline in FEV1 in COPD is related to thickeningof thewalls of thesesmall conductingairways (4). Thereisalsoevidencethatthepathophysiologicpathwaysthatleadtoemphysemaandtosmall airwaynarrowingareindependent;individual patients mayhavepredominantlysmall airwayorparenchymal disease. Recently, Nakano and colleagues (7)showedthat anincreaseinthicknessof theapical segmentalbronchusin therightupper lobemeasured onhigh-resolutioncomputedtomography(HRCT)wasrelatedtotheseverityofairowobstructionandgastrapping(FEV1, FVC, andFEV1/FVC) in smokers who had COPD independently of the degreeof emphysema measured on the same HRCT scans. Unlike thedegree of emphysema, the HRCT airway dimensions were notrelated to the diffusing capacity. Because the important site ofairway narrowing in COPD is the bronchioles, these data suggestthatthickeningoflargeairwaysmaybeasurrogateforsmallairway abnormalities. It is possible that the same pathophysio-logicprocess, whichresults inexcessiveobstructionof smallairways, also occurs in the larger airways (8). Although increasedairway wall thickness in the larger airways may have little func-tional consequence, the ability to measure thickening using CTcould prove to be a useful predictor of small airway pathologyand may allow for phenotypic stratication of patients who haveCOPDintoparenchymal-andairway-predominantcategories.To test this hypothesis, we measured the airway dimensions oflarge- andintermediate-sizedairwaysusingHRCTandcom-paredthesemeasurements withhistologicestimates of smallairway remodeling in resected human lungs. Some of the resultsof this study have been previously reported in an abstract (9).METHODSThis study was part of an ongoing investigation of lung structure andfunction (10). Twenty-two patients gave informed consent to have theirlungfunction,computedtomography(CT)scans,andresectedtissueexamined in research studies using methods approved by the ProvidenceHealth Care Clinical Ethics Review Board. The subjects were consecu-tivepatients whorequiredpneumonectomyor lobectomyfor smallperipheral lungnodules; theywerenot selectedtorepresentanob-structed and nonobstructed group.Before surgery, subdivisions of lung volume, spirometry, and single-breath diffusing capacity were measured as previously described, andthey conformed to American Thoracic Society standards (1012). Im-mediately after resection, the lung or lobe was obtained from the op-erating room, inated to a transpulmonary pressure of 20 cm H2O, andpositioned in either a GE CT/i or a High-Speed Advantage CT scanner(General Electric Medical Systems, Milwaukee, WI) in an orientationsimilar to a clinical supine CT scan. Axial images were acquired usingeither 1.0- or 1.5-mm collimation and were reconstructed using a highspatial frequency reconstruction algorithm.Aftercompletionof theCTscan, theresectedlungorlobewasprepared for histology as previously described (13) and detailed in theonline supplement. Images of all intact membranous and cartilaginousairways cutin reasonable cross-section(long/short diameter of3.3 orless) were captured using a digital camera at an appropriate magnica-tion (14). The digital images of the airways were analyzed using ImageNakano, Wong, de Jong, et al.: CT Airway Dimensions 143Pro Plus software (Media Cybernetics, Silver Spring, MD). The luminalperimeter(Pi)wasmanuallytraced,aswastheadventitialperimeterat theborder betweentheairwaywall andlungparenchyma. Theperimetersoftheairwaylumenandoftheadventitiasubtendedtwoareas: Ai (luminal area) and Ao (total area). Airway wall area (Aaw)was calculated as Ao Ai. An estimate of small airway remodeling wasderivedforeachpatientbyplottingPiagainstthesquarerootofwallarea (15) using linear regression analysis for each subject and calculatingeach individuals value of Aaw at a Pi of 4 mm (Figure 1).All airways visibleonCTandcut inareasonablecross-section(long/short internal diameter of 2.2 or less) were analyzed. Every fthCT slice was used to avoid the possibility of overlap. Airway dimensionswere measured using customsoftware written for the numerical analysispackage PV-Wave (Visual Numerics, Boulder, CO) as previously de-scribed (14) and as detailed in the online supplement. Briey, 64 rayswere projected from a seed point in the lumen, and the X-ray attenua-tion values were measured along each ray. The Aaw was dened usingthe full width at half maximum principle. Manual editing was used toremoveraysthatprojectedbeyondtheairwaywallintoneighboringdense structures such as pulmonary arteries (Figure 2). In addition tothe determination of Ai, Ao, and Aaw (Ao Ai), we also calculatedthe wall area percentage (WA%) as Ao Ai/Ao 100.ThedifferencebetweentheslopesandinterceptsobtainedfromhistologyandCTwas testedusingtheStudents t test. Regressionanalysis was performed to test the relationship between the slopes ofthe Pi versus Aaw determined from CT and histology for each case,as well as to test the relationship between WA% assessed by CT withthe estimateof smallAaw measuredhistologically (Aawat aPi4mm [internaldiameterof1.27 mm]).Dataareshown asmeanSD. A p value of less than 0.05 was considered signicant.RESULTSThepulmonaryfunctionof thepatientsisshowninTable1.Five of the 22 subjects were nonsmokers, and 1 had smoked foronly 1 year. These data show that there is a wide range in thelevel of obstruction, but most had only mild obstruction (GlobalInitiative for Obstructive Lung Disease [GOLD] classes 1 and 2)(16). Figure1showstherelationshipbetweenPi andAawmeasured by CT and histology for one patient. Examination ofFigure1illustratesthatthesize(Pi)oftheairwaysmeasuredFigure 1. Case 6094: The relationships between airway internal perime-ter (Pi) andairwaywall area(Aaw) measuredhistologically(opensquares) and by computed tomography (CT) (closed circles). It is appar-ent that CT markedly overestimates Aaw, especially in smaller airways.Thelongvertical dashedlineindicatesthePi cutoffthatwasusedforCT(0.75cm).Themeanwallareapercentage(WA%)forallairwaysmore than this threshold (encircled by a solid line) was compared withthehistologicmeasurementof AawataPi of4mm(shortverticaldashedline). Adownward-facingarrowreferencesthepointatwhichairways are at a diameter of 2 mm.using CT is larger than the membranous and cartilaginous air-ways assessed by histologic examination, although there is someoverlap. It is also apparent that the Aaw at any Pi is substantiallygreater as measuredby CTthanonhistologic examination. Therewas a small signicant difference in the slopes of Aaw versusPi (0.23 0.03 vs. 0.28 0.09, p 0.002) measured by CT andhistology, respectively. However, therewasalargesignicantdifferenceintheintercepts(0.150.01vs. 0.020.01, p0.0001). TherewasalsoasignicantrelationshipbetweentheslopesmeasuredusinghistologyandCT(i.e., individualswhohad a steep slope of Pi vs. Aaw on histology also had a steepslope on CT) (CT slope 0.2059 histology slope 0.1701, R20.32, p 0.01; Figure 3). A weak but signicant correlation wasalsofoundfor therelationshipbetweenintercepts (R20.13,p 0.05).Because we have previously shown (14) that there is an over-estimation of Aaw on CT and that this overestimation is greatestinsmall airways, weperformedananalysis restrictedtothelarger airways on CT. We calculated the WA% for all airwayswithanPiofgreaterthan0.75 cmmeasuredonCT(meanSD Pi 1.036 0.142cm) and compared thisvalue with thehistologic estimate of small airway remodeling (Aaw at a Pi 4 mm). We chose the value of Aaw at a Pi 4 mm becausethis conforms to airways with an internal diameter of approxi-mately 1.27 mm, which is the size of airways that has been showntobethesiteof increasedresistanceinCOPD(2). Figure4shows thesignicant relationshipbetweentheCTmeasuredWA%ofall airwayswithaPiofmorethan0.75cmandthepredicted AawforairwayswithaPiof4mmmeasuredonhistology (R20.57, p 0.01). This relationship was also signi-cant when we used a CT Pi cutoff of 1 cm (R2 0.39, p 0.01).DISCUSSIONThe results of this study indicate that the airway dimensions ofthelarge-andintermediate-sizedairways, whichcanbeaccu-rately assessed using HRCT scanning, reect airway dimensionsinthesmaller airways, whicharethemost important siteofairway obstruction in COPD.Two pathophysiologic processes contribute to the pathogene-sisofCOPD: proteolyticdestructionofthelungsconnectivetissueframework, whichcausesemphysemaandloss of lungrecoil and inammatory/brotic narrowing of peripheral airways(4, 17, 18). Both of these processes result in similar physiologicderangements, includingincreasedairwayresistance, decreasedmaximal expiratoryow, hyperination, gas trapping, andgasexchange impairment. It is possible that different genetic suscepti-bility, interacting with environmental factors, favors the predomi-nant development of one of these pathophysiologic processes inindividual smokers. The work of Nakano and colleagues (7) sug-gested that these phenotypes can be separated by measuring thepercentage of low attenuation area in the lung as a marker ofemphysema and the dimensions of the apical segmental bronchusto the right upper lobe as an estimate of airway remodeling. Hefound that the percentage of low attenuation area and relativeAaw(WA%)independentlycontributedtothepredictionofFEV1, FVC, and FEV1/FVC but that only the percentage of lowattenuation area was related to diffusing capacity.However, the apical segmental bronchus is a large cartilagi-nousairway, andnarrowingofsuchairwaysisnotthoughttocontribute signicantly to increased airway resistance in patientswho have COPD. It has been shown (2) that the major site ofairwayobstructioninpatientswhohaveCOPDisinairwayssmaller than 2-mm internal diameter (Pi 6 mm), and thisnding has been conrmed (5, 6). On the other hand, Tiddensand colleagues (8) showed that histologic estimates of cartilagi-144 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 171 2005Figure 2. Airway measurement (14)(also detailed in online supplement). (A)A seed point is placed in the lumen of amagnified airway, and the X-ray attenua-tionvaluesaremeasuredalong64raysprojected from the seed point. The Aawisdefinedusingthefull widthat halfmaximum principle of each ray, and thelumen and wall perimeter are measuredby connecting the endpoints of the lines.(B) Manual editing of rays is used to re-move rays that project beyondthe airwaywall intoneighboringdensestructuressuch as pulmonary arteries.nous airway dimensions were related to airow obstruction andto peripheral airway inammatory scores. We reasoned that theairway wall remodeling and luminal narrowing might involve alllevelsofthe tracheobronchialtreeinsusceptible smokersandbedetectableusingHRCT. Althoughitisthethickeningandnarrowing of the peripheral airways that leads to the most impor-tant functional consequences, thickening of larger airways couldprovideasurrogatemeasureofthesmallairwayprocess.Theresults of this study support this hypothesis and extend the obser-vations of Nakano and colleagues (7).Wehavepreviouslydemonstratedthatthefractional errorin estimation of airway wall and lumen area increases in smallairways (14, 19). This error is caused by a systematic overestima-tionofwall areaandunderestimationofAi. Thereasonsforthe systematic error include but are not limited to the following:(1) Volume averaging: airways do not always run perpendicularto the scanning plane of the CT and even with a slice thicknessof only 1 mm, angling of the airway can lead to signicant error(15). (2) The mucosal folds in an air inated lung are lled withliquid (20) so that the potential lumen area in the folds cannotbe appreciated on CT scans but is included in the determinationof maximal lumenareaonhistologicassessment. (3)Airwaysmooth muscle relaxation induced by xation after the scanningmay cause a true increase in Ai. (4) The edge detection algorithmof our method is affected by the limited spatial resolution andpoint spread function of the CT scanner. (5) Shrinkage of histo-logic specimen during the xation process will cause and appar-ent decrease in Aaw.Overestimation of Aaw measurements can be caused by tan-gential sectioning of the airways both on histologic examinationand on CT. To minimize this source of error, we limited to theanalysis to airways that were cut in reasonable cross-sectionbasedonthelongtoshortdiameterratio. Weuseddifferentlong:shortdiameterratiosforCT(2.2:1)andhistology(3.3:1)basedonpreviouslypublishedcriteriaforairwayselection(4,TABLE 1. PULMONARYFUNCTIONDATAMean (Median) Range SDAge 64 3574 9Male/female 11/11FEV1% predicted 83 (85) 55114 16FVC % predicted 92 (96) 56122 17FEV1/FVC % 71 (74) 4989 10DLCO % predicted 78 (75) 43130 23Smoking pack-years 29 (18) 090 25Definition of abbreviation: DLCO carbon monoxide diffusing capacity.14). The smaller ratio used for the CTanalysis reects the greaterpotential for overestimation of wall area using this method be-cause of volume averaging over the 1- or 1.5-mm slice thickness.Althoughtheuseof thedifferent ratiocutoffscouldleadtoslight differences in the wall areas estimated by the two methods,we do not think it is a signicant source of discrepancy or wouldaffect our major conclusions.The overestimation of Aaw by CT is conrmed in our data, asillustratedinFigure1; thefractionaldifferencebetween AawmeasuredbyCTandhistologyisgreatestinsmall airwaysanddiminishes in larger airways. Despite this overestimation, Figure 3illustrates that there is a signicant correlation between CT andhistology slopes of Pi versus wall area, and we also found a weakbut signicant correlation of the intercepts. These results suggestthat despite systematic overestimation of wall area, CT airwaydimensions reect histologic airway dimensions. Furthermore, acomparison of CTand histology using CTmeasurements derivedonlyfromthelargerairwaysshowedthatthemeanWA%inairways with a Pi of approximately 1 cm on CT was correlatedwith the histological estimates of wall area in airways with a Piof 4mm. Welimited thisanalysis tothe largerairways astherelationshipbetweenPiandAawonCTisheavilyinuencedbythemorenumeroussmaller airwaysinwhichAawisover-estimatedforthereasonsdiscussedpreviouslyhere. Inaddi-tion, WA% for the more accurately measured airways ( 0.75and1.0cm)isausefulsummaryvaluebecauseitprovidesaneasily understood and implemented pooling of the results fromFigure 3. Comparisonof slopes: The slope of Pi versus Aawonhistologyisplottedagainst theslopeof Aaw(y0.2059x0.1701, R20.3206, p 0.01).Nakano, Wong, de Jong, et al.: CT Airway Dimensions 145Figure 4. Comparison of CT WA% versus histology wall area: individualsubjects mean WA% for all airways with a Pi of 0.75 cm or greater asassessedonCT versus the value for Aawat a Pi of 4 mmderivedfromtheindividual relationships between Pi and Aaw measured histologically.different sizedairways. Wealsoanalyzedthewall areaat aspecic Pi on CT and found a signicant although less powerfulassociation with the histologic estimates (data not shown).TheCTscansof thelobesandlungswereobtainedafterresection. Thisallowedprecisecontrol ofthetranspulmonarypressure to which the lobes were inated and complete absenceof movement artifact related to breathing or cardiac motion. Inaddition, the scans were obtained using a eld of view of 20 cminsteadofthe3540cmusedduringinvivoscans.Therefore,thevoxel dimensions inthis studywereapproximatelyone-fourthofthevoxel sizeonmostclinical CTscans, indicatingthatCTimagesmayneedtobereconstructedusingaeldofview of 20 cm to attain the same precision as this study.The aim of this study was not to relate lung function to airwaydimensions because the number of subjects was relatively smalland their degree of airway obstruction relatively mild. Althoughthere were trends for a relationship between histologic and CTairwaydimensions andFEV1andFEV1/FVC, thesedidnotachieve statistical signicance in this small group. Previous inves-tigators have shown signicant relationships between lung func-tionandhistologic(4, 18)andCT(7)airwaydimensions, butthis is the rst study in which histologic and CT dimensions havebeen compared in the same subjects. These results respond tothe recent appeal for pathologic validation of CT measurementsin COPD (21). It will be important to validate these ndings inpatients who have more severe disease.Thereisgrowingawarenessthatanytherapeuticinterven-tions designed to prevent the brosis and scarring found in pe-ripheral airways in COPD are likely to be different from thosedesignedtoattenuatetheemphysematous destructionof thelung parenchyma. Therefore, it is important to develop in vivotechniques to quantify the contribution of small airway obstruc-tion and emphysema in the lungs of patients who have COPD.Such techniques are required to both stratify patients in clinicaltrials and tailor specic therapeutic agents to the patients pre-dominant pathophysiologicprocess. Theresults of this studysuggest that HRCT can be used for these purposes.Conflict of Interest Statement : Y.N. does not have a financial relationship with acommercial entitythat has aninterest in thesubject of thismanuscript; J.C.W.does not have a financial relationship with a commercial entity that has an interestin the subject of this manuscript; P.A.d.J. does not have a financial relationshipwith a commercial entity that has an interest in the subject of this manuscript;L.B. does not have a financial relationship with a commercial entity that has aninterest in the subject of this manuscript; T.N. does not have a financial relationshipwith a commercial entity that has an interest in the subject of this manuscript;H.O.C. has received$2,500in2002and1,500on2003for servingonanadvisory board for GlaxoSmithKline (GSK), is a co-investigator of two multicenterstudies sponsored by GSK, and has received travel expenses to attend meetingsrelatedtotheproject,andapercentageof hissalarybetween2003and2006($15,000/year) derives from contract fund provided to a colleague Peter D. Pareby GSK for the development of validated methods to measure emphysema andairwaydiseaseusingCT; W.M.E. doesnothaveafinancial relationshipwithacommercial entitythathasaninterestinthesubjectofthismanuscript;J.C.H.has received funding fromGSK to sponsor part of this study; P.D.P. is the principalinvestigator of a project funded by GSK to develop CT-based algorithms to quanti-tate emphysema and airway disease in COPD, and with collaborators has received$300,000 to develop and validate these techniques.Acknowledgment : The authors acknowledge Kenneth P. Whittall, Ph.D., and Anh-Toan Tran, B.Sc., for technical assistance in developing and supporting the airwayanalysis application.References1. Cosio M, Ghezzo H, Hogg JC, Corbin R, Loveland M, Dosman J, Mack-lemPT. Therelationsbetweenstructural changesinsmall airwaysand pulmonary-function tests. N Engl J Med 1978;298:12771281.2. Hogg JC, MacklemPT, Thurlbeck WM. Site and nature of airway obstruc-tion in chronic obstructive lung disease. N Engl J Med 1968;278:13551360.3. Cosio MG, Guerassimov A. Chronic obstructive pulmonary disease: in-ammation of small airways and lung parenchyma. Am J Respir CritCare Med 1999;160:S21S25.4. Hogg JC, Chu F, Utokaparch S, Woods R, Elliott WM, Buzatu L, Cherni-ackRM,RogersRM,SciurbaFC, CoxsonHO,etal.Thenatureofsmall-airwayobstructioninchronicobstructivepulmonarydisease.N Engl J Med 2004;350:26452653.5. VanBrabandtH, CauberghsM,VerbekenE, MoermanP,LauwerynsJM,VandeWoestijneKP. Partitioningofpulmonaryimpedanceinexcised human and canine lungs. J Appl Physiol 1983;55:17331742.6. Yanai M, Sekizawa K, Ohrui T, Sasaki H, Takishima T. Site of airwayobstructioninpulmonarydisease: directmeasurementofintrabron-chial pressure. J Appl Physiol 1992;72:10161023.7. NakanoY, MuroS, SakaiH, HiraiT, ChinK, TsukinoM, NishimuraK, Itoh H, ParePD, Hogg JC, et al. Computed tomographic measure-ments of airway dimensions and emphysema in smokers: correlationwith lung function. Am J Respir Crit Care Med 2000;162:11021108.8. Tiddens HA, Pare PD, Hogg JC, Hop WC, Lambert R, de Jongste JC.Cartilaginous airwaydimensions andairowobstructioninhumanlungs. Am J Respir Crit Care Med 1995;152:260266.9. Nakano Y, Buzatu L, Wong JC, de Jong PA, Elliott WM, Whittall KP,Hogg JC, Pare PD. Small airway dimensions in human lungs: compari-son of CT to histology [abstract]. Am J Respir Crit Care Med 2003;167:A585.10. Hogg JC, Wright JL, Wiggs BR, Coxson HO, Opazo Saez A, Pare PD.Lung structure and function in cigarette smokers. Thorax 1994;49:473478.11. Coxson HO, Hogg JC, Mayo JR, Behzad H, Whittall KP, Schwartz DA,Hartley PG, Galvin JR, Wilson JS, Hunninghake GW. Quanticationof idiopathic pulmonary brosis using computed tomography and his-tology. Am J Respir Crit Care Med 1997;155:16491656.12. Coxson HO, Mayo JR, Behzad H, Moore BJ, Verburgt LM, Staples CA,Pare PD,HoggJC.Measurementoflungexpansionwithcomputedtomography and comparison with quantitative histology. J Appl Phys-iol 1995;79:15251530.13. Coxson HO, Rogers RM, Whittall KP, DYachkova Y, Pare PD, SciurbaFC, Hogg JC. A quantication of the lung surface area in emphysemausing computed tomography. Am J Respir Crit Care Med 1999;159:851856.14. NakanoY, Whittall KP, KallogerSE, CoxsonHO, Flint J, Pare PD,EnglishJC. Development andvalidationof humanairwayanalysisalgorithmusingmultidetectorrowCT. In: CloughAV, ChenC-T,editors. Medical imaging 2002: physiology and function from multidi-mensional images. ProceedingsofSPIETheInternational Societyfor Optical Engineering, Vol. 4683. SPIE; 2002. pp. 460469.15. Kuwano K, Bosken CH, Pare PD, Bai TR, Wiggs BR, Hogg JC. Smallairways dimensionsin asthmaand inchronic obstructivepulmonarydisease. Am Rev Respir Dis 1993;148:12201225.16. Pauwels RA, Buist AS, Calverley PM, Jenkins CR, Hurd SS. The GOLDScientic Committee: NHLBI/WHO Global Initiative for Chronic Ob-structiveLungDisease(GOLD)Workshopsummary. AmJRespirCrit Care Med 2001;163:12561276.17. Cosio MG, Hale KA, Niewoehner DE. Morphologic and morphometriceffects of prolonged cigarette smoking on the small airways. Am RevRespir Dis 1980;122:265271.146 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 171 200518. Bosken CH, Wiggs BR, Pare PD, Hogg JC. Small airway dimensions insmokers with obstruction to airow. Am Rev Respir Dis 1990;142:563570.19. KingGG, MullerNL, Pare PD. Evaluationof airwaysinobstructivepulmonary disease using high-resolution computed tomography. AmJ Respir Crit Care Med 1999;159:9921004.20. Yager D, Butler JP, Bastacky J, Israel E, Smith G, Drazen JM. Ampli-cation of airway constriction due to liquid lling of airway interstices.J Appl Physiol 1989;66:28732884.21. CroxtonTL, WeinmannGG, SeniorRM, Wise, CrapoJD, BuistAS.Clinical research in chronic obstructive pulmonary disease: needs andopportunities. Am J Respir Crit Care Med 2003;167:11421149.