estimation lung · and helium-dilution estimates of total lung capacity (tlc), taking account of...

Thorax, 1979, 34, 726-734

Estimation of lung volumes from chest radiographsusing shape informationR J PIERCE, D J BROWN, M HOLMES, G CUMMING, AND D M DENISON

From the Lung Function Unit, Brompton Hospital, London, andMidhurst Medical Research Institute, Sussex, UK

ABSTRACT The cross-sectional shapes of the chest and its contained structures have been assessedin post-mortem anatomical sections and from computerised tomographic scans in livingsubjects. These shapes are described by simple equations that can be used to increase theaccuracy of measuring lung volumes from chest radiographs. Radiographic estimates of totallung capacity, using the equations, were compared with plethysmographic and single-breathhelium dilution measurements in 35 normal subjects. The postures commonly used for takingchest radiographs were found, on average, to decrease total lung capacity (TLC) and to increaseresidual volume by about 200 ml when compared with the sitting positions used for the othertwo measurements (studies made in 18 of the subjects). After correction for this effect, theradiographic estimates of TLC, which measure the displacement volume of the lung, exceededthe plethysmographic estimates of contained gas volume by a mean of 720 ml, which was

taken as the volume of tissue, blood, and water in the lungs. The single-breath dilution estimatesof TLC fell short of the plethysmographic values by a mean of 480 ml, taken as the volume ofcontained gas that was inaccessible to helium in 10 seconds. The tomographic studies suggestedthat the radiographic technique of measuring lung displacement volumes has an accuracy of+210 ml. The method is rapid and simple to use and has intra- and inter-observer variabilitiesof <1% and <5% respectively.

Radiographic methods for measuring lung volumehave been known for many years. Most have beenbased on simple assumptions about the cross-sectional shape of the chest. Hurtado and Fray(1933) measured the area of the lungs on postero-anterior (PA) chest radiographs and multiplied thisby the PA diameter of the chest measured exter-nally. Kovach et a! (1956) treated the chest as aparaboloid of revolution of the PA radiograph.Barnhard et al (1960) assumed that each lung andthe heart were elliptical in cross-section and con-sidered the diaphragm as one-eighth of anellipsoid. All these methods are dependent for theiraccuracy on correlations with other methods ofmeasuring lung volume, such as plethysmographyor gas dilution. The acceptability of the plethys-mograph as a standard for comparison is notentirely above suspicion, especially in patients withlung disease: recent studies by Brown et al (1978b)have suggested that airways obstruction may leadto inaccurate measurement. Previous studies havealso neglected the effects of postural differences on

static lung volumes, and have applied a singleapproximate magnification factor to all radio-graphs.The radiographic method described herein

measures the displacement volumes of the lungsas geometric structures in space, using a digitiserand computer. It was developed from determina-tions of the cross-sectional shape of the chest andits contained structures in post-mortem anatomicalsections and from computerised tomographic (CT)scans in living subjects. It also uses individualmagnification factors that are easily obtained atthe time of measurement. We describe the methodand compare its results with plethysmographicand helium-dilution estimates of total lungcapacity (TLC), taking account of postural effectson thoracic gas volume. In principle the radio-graphic method estimates the displacement volumeof the lung, the plethysmographic technique deter-mines the contained gas volume, and the dilutionmethod represents the volume of gas reached byhelium in the duration of the breath (10 s). Thus

726

copyright. on M

arch 19, 2020 by guest. Protected by

http://thorax.bmj.com

/T

horax: first published as 10.1136/thx.34.6.726 on 1 Decem

ber 1979. Dow

nloaded from

http://thorax.bmj.com/

Radiographic lung volume

the differences between these three measurementsare estimates of lung tissue volume and effectivegas trapping respectively.

Methods

RADIOGRAPHIC METHODThe principles of the radiographic method can bethought of in four stages (fig 1). Firstly, theboundaries of the chest, heart, spine, anddiaphragm are identified on PA and lateral radio-graphs and traced over using a digitiser that feedsthe XY co-ordinates of successive points of eachoutline directly into a computer. Secondly, the

computer aligns the PA and lateral views in thesame vertical plane, divides them into a largenumber (200) of horizontal slices, and determinesthe PA and lateral diameters and thickness ofeach structure in the slice. Thirdly, the cross-sectional areas of the structures are calculatedfor each slice, using shape equations describedbelow. Finally, the volumes of the whole organsare calculated by summing information from all ofthe slices and the results are presented in graphicand digital form.The outlines used illustrated in fig 2 are re-

corded using a Summagraphics digitiser with aresolution of 100 lines per inch and a Prime 300

Principle

Chest volumeminus subphrenic volumernin us heort volumeminus spine volume

lung volume

Metho d

(i) Sketch round the PA and loteroloutline shown

(ii) Divide outline into mony horizontolslices

(Nii) Assume they are ellipses normolto film plones

(iv)Colculote volume of slices ond sum(digitol plotting routine tokes 2minper poir of tilms)

Fig 1 General principles of radiographic method.

Fig 2 PA and left lateral radiographs of a normal subject showing outlines that are digitised to estimatevolumes of each of the thoracic structures.

727

copyright. on M



/T


ber 1979. Dow

nloaded from


R J Pierce, D J Brown, M Holmes, G Cumming, and D M Denison

computer. In the PA radiograph the chest outlinefollows the inner rib margins, the heart boundaryincludes the arch of the aorta, and the spineoutline follows the tips of the lateral processes ofthe vertebrae to take account of associated musclemasses. In the lateral view the chest boundaryfollows the inner border of the sternum anteriorlyand the inner rib margins posteriorly. Where rota-tion of the chest displaces the two rib marginsposteriorly a line midway between the two istaken. Similarly a line midway between the twohemidiaphragms is followed. The heart outlineagain includes the arch of the aorta as farposteriorly as the anterior limit of the spinal mass.This is drawn 1 cm in front of the vertebral bodiesto allow for the oesophagus and descending aorta.Posteriorly the boundary of the spine follows thatof the chest. We normally use the top of theaortic arch as a common point of reference forvertical alignment but any other structure that canbe clearly identified in both views serves equallywell.The individual magnification factor for each

radiograph was estimated from the target-filmdistance and half the chest diameter normal tothe film in that view (fig 3). All radiographs usedin this study were taken at a distance of 10 ft, butthe calculation is straightforward for any cir-cumstances in which the geometry is known.Where this is not practical the magnificationfactor can be obtained simply and directly bytaping coins to the back, front, and sides of thechest and noting their maximum diameters as seenon the radiographs. The precise positions of thecoins are not important since it is impossible toplace any disc in a field of view without its truediameter being visible.The accuracy of the method was assessed by

taking PA and lateral radiographs of several con-

-t-

Mag - Yx

ta

-g-

tainers, of known cross-sectional shape, understandard chest film conditions. The radiographicvolumes were compared with the containers'water displacement volumes, which varied from0 25 to 14-35 1 and thus covered the range ofchest volumes to be expected in clinical practice.

ASSESSMENT OF THE CROSS-SECTIONAL SHAPE OF

THE CHEST IN MAN

The cross-sectional shapes of the chest and itscontained structures were determined fromanatomical and tomographic material. Theanatomical data were obtained from serial frozensections of the whole chest reported by Matsukawaet al (1977) and Eycleshymer and Schoemaker(1970). The areas of the chest, heart, spine, andsubdiaphragmatic structures in each slice weremeasured by planimetry using the digitiser andcomputer mentioned previously. The PA andlateral diameters of each structure, appropriateto the radiological outlines described earlier, werethen measured as illustrated (fig 4). The truecross-sectional area of each structure in the slicewas then compared with the areas of the ellipseand rectangle generated by the same two diam-eters. The volume of the whole structure wascalculated using the slice thickness and photo-graphic magnification factors stated by theauthors.

Serial CT scans were obtained of the chests offour normal male subjects (aged 21-54) duringbreath-holding at TLC. Three further people werestudied during breath-holding at functionalresidual capacity (FRC)-a man aged 37 with nohistory of lung disease, a woman of 38 withasthma, and a man of 75 with emphysema. Threenormal men including two from the TLC groupwere studied at residual volume (RV). The samelinear and planimetric measurements were made

IiFig 3 Principles of estimationof magnification factors fromtarget-film distance (t), air gapbetween subject and film (g),and half chest diametermeasured from radiographtaken in other projection ordirectly from patient's chest.This magnification factor appliesto plane at centre of chest.

728

copyright. on M



/T


ber 1979. Dow

nloaded from



(Adapted from Matsukawa etal 1977 )

Fig 4 Cross-sectional areas measured by planimetryand the PA and lateral linear dimensions of eachstructure. By determining relation between thesemeasurements, average cross-sectional shape of eachstructure was established.

on the tomographic as on the anatomical sections.Once more actual volumes were calculated fromknown magnifications and "slice" thicknesses. AllCT scans had to be conducted with the subjectssupine.

MEASUREMENTS OF LUNG VOLUMESLung volumes were measured in three ways:

radiography, whole-body plethysmography, andsingle-breath helium dilution. Studies were con

ducted on 35 normal adults, 17 of whom were

women. Their nmean age was 28-9 years (range18-52), their mean height 174-6 cm (range 152-188), and their mean weight 67-1 kg (range 45-98).Radiographic estimates of lung volume-

Standard 10 ft-150 KeV chest radiographs were

taken in the PA and left lateral positions. In thePA position subjects stood with hands on hipsand shoulders pressed forward on the film cassette.For the lateral view subjects stood upright witharms clasped tightly about the head and with theleft side of the chest against the cassette. Thesubjects were told of the importance of inspiring

completely to TLC and signalled when they hadreached this point by pressing a toy "clicker"that was audible to the radiographer behind herscreen. The cassette was fitted with a constantsubject-film air gap of 5 cm in both views.

Whole-body plethysmography-For thesemeasurements subjects sat upright in a constant-volume whole-body plethysmograph (Fenyves-Gut,Basle). Thoracic gas volumes were determinedfrom inspiratory efforts against a shutter closedat FRC. TLC was obtained from the maximalinspiration that followed release of the shutter.The mean of at least three reproducible measure-ments was obtained in each subject. Calculatedvolumes were reduced by 130 ml in all cases toallow for abdominal gas compression (Brown et al,1978a; Habib and Engel, 1978).

Single-breath helium dilution-For thismanoeuvre subjects sat upright, exactly as in theplethysmograph, expired to RV, rapidly inspired ahelium-rich mixture to TLC, held their breath for10 seconds, and then expired quickly back to RVprecisely as in the standard single-breath DLCOprocedure. The helium concentration of the mid-expirate (750-1250 ml expired volume) wasmeasured with a P K Morgan Mk IV respara-meter. The mean of at least two reproduciblemeasurements was obtained in each subject. Thesingle-breath, plethysmographic, and radiographicestimates of TLC were made within a one-hourperiod in each person.

Postural changes in TLC and RV-The changesin TLC and RV between the standing posture ofthe radiographic procedure and the sitting positionof the other two were determined in 18 of the 35subjects. This was done by connecting a water-filled spirometer to the subject, who was in theposition adopted for the PA radiograph, and ask-ing him to perform a vital capacity manoeuvre inthat posture and then to repeat the manoeuvre inthe normal sitting position, without coming off themouthpiece. This procedure was repeated using thelateral radiograph position. Duplicate measure-ments were obtained in reverse order on eachsubject and the main effects of posture upon TLCand RV calculated. A typical record illustratingthe procedure is shown in fig 5.

Observer error-To assess the accuracy withwhich lung volumes can be measured from a givenpair of radiographs one observer measured 18 setsof radiographs in triplicate and three observersmeasured five sets of films in triplicate. Two ofthe three had no previous experience of themethod.

729

copyright. on M



/T


ber 1979. Dow

nloaded from



RP Normal

TLC

RV UPA CXR Rebxed

Relaxed

LAT CXR

i,1I! ,j!l

\J.

PA CXR

Relaxed LAT CXR

Results

CROSS-SECTIONAL SHAPE ASSESSMENT

The average cross-sectional shape of the chest inthe anatomical studies exceeded that of the ellipse,but was smaller than that of the rectangle gener-

ated by the PA and lateral linear dimensions. Thearea was found to be one-third of the way be-tween the elliptical and rectangular shapes (fig 6),

and is described by the equation:

A=7

+(ab- ab)/3,which simplifies to 0-857 ab, where a and b are

the linear dimensions. Table 1 shows the error inestimating the actual volume of the chest fromits PA and lateral diameters using this cross-

sectional shape for both the anatomical sectionsand the CT scans. The mean error for the 10 CT

Ellipse area = obRectangular area = ab

Fig 6 Diagramatic representation showing averagecross-sectional area of chest (solid line follows pleuralmargin). Area of chest was found to be greater thanthat of ellipse based on PA and lateral lineardimensions (inner broken line) by one-third ofdifference between area of ellipse and that of enclosingrectangle (outer broken line).

Table 1 Comparison of estimated with actual volumeof the whole chest

Actual Estimated* Error ofchest chest estimatevolume (1) volume (l) (%)

Anatomical sectionsMatsukawa et al (1977) 4-69 4-68 -0-2Eycleshymer and Schoemaker 6-56 6-49 -1 0(1970)

CT scansTLC-normal subjects

1 10-67 10-46 -2-02 8*55 8*53 -0-33 12-64 13-04 +3-14 1161 11-14 -4*1

FRC-patients5 Normalchest 7-88 8 01 +1*66 Asthma 6-83 6-92 +0 57 Emphysema 9*24 9-29 +0 5

RV-normal subiects3 7-69 7-68 -0014 5*63 5*42 -3-78 6*34 6*27 -1.1

Mean error for the 10 CT scans= -05 % (SD 2-2%).,vab -nab

3*Estimatebased on: Cross-sectional area= +(ab --)/3where a and b are PA and lateral diameters ofthe chest.

Fig 5 A typical spirographic record showing effectof radiographic postures in reducing TLC andincreasing RV.

I

730

copyright. on M



/T


ber 1979. Dow

nloaded from



scans was -0 5% of a mean chest volume of8-71 1 (ie 45 ml), and the SD of the error was2 2% (190 ml).The average cross-sectional areas of the heart,

spine, and subdiaphragmatic region were found tobe very close to those of the ellipses generated bytheir PA and lateral diameters. The errors inherentin the estimation of these volumes from ana-tomical and tomographic data, by making thisassumption, are shown in table 2. The means andstandard deviations of these errors are all small(<50 ml). By compiling the mean errors for thechest, heart, spine, and subphrenic volumes fromthe CT scan data, the total lung volume wasfound to be underestimated by 125 ml. Combiningthe variances yields an overall accuracy for theestimation of lung volume of +i=210 ml. Lungvolume is calculated from the other volumes so:

Lung vol=vols chest-(heart+spine+ subphrenum)+ 125 ml

Only two subjects had CT scans at both TLCand RV. In both there was little change in theshape of the chest between the two volumes. Thissuggests that the expansion of the chest isrelatively isotropic over the vital capacity range.The cross-sectional shape of the heart, sub-diaphragmatic structures and, more predictably,the spine also showed little change between TLCand RV.The results of the comparison of radiological

volume and water displacement volume of thecontainers are shown (fig 7). Agreement betweenthe two measurements is extremely close.The study of the effects of posture on lung

volumes showed that TLC was reduced by a meanof 0-71 1 (SD 0 16 1) in the PA radiographicposition when compared to the sitting relaxedposition, and by 018 1 (SD 0 12 1) in the lateralradiographic position. Residual volume, on theother hand, was increased by 0-20 1 (SD 013 1) in

Table 2 Elliptical estimation of heart, spine, andsubdiaphragmatic volumes

Actual Error ofvolume (l) estimate

(%)Anatomical sources (n =2)

Heart 0O77±0O13 -0O6±5-9Spine 062±012 -3 0±0±1Subdiaphragmatic volume 133 ±0O49 +0-4±2-9

CT scans (n= 10)Heart 101±+0 34 +2-8+4-7Spine 0 88±0 15 +06+5-5Subdiaphragmatic volume 176± 1-06 +2-8±+2-3

Estimates based on: Cross-sectional area=wab/4 where a and b are PAand lateral linear dimensions ofeach structure. Figures are mean± I SD.

212-

E

>

04E4

1-

O 1 4 8 12 16Displacement volume I

Fig 7 Correlation of radiographically determined andwater displacement volumes for series of containersused to evaluate accuracy of radiographic method.Line is line of identity (slope= 1JO).

the PA position, and by 0-21 1 (SD 0O14 1) in thelateral position. Accordingly, in the comparisonbetween radiological TLC and the other volumeestimates described below, 0 18 1 has been addedto the radiological volume, to correct for thepostural differences.

LUNG VOLUME COMPARISONSThe mean values for the three lung volumes andthe differences between them are shown in table 3.The correlations between radiographic andplethysmographic and helium dilution TLC areshown in figs 8, 9, and 10 respectively. Thecorrelation coefficients were 0-96, 0-95, and 0-97respectively and were all highly significantstatistically (P<0-01, method of least squares).The mean difference between radiographic andplethysmographic TLC was +0-72 1 (SD 0-37 1) andis attributable to lung tissue, water, and bloodvolumes. The mean difference between radiographicTLC and helium dilution volume was + 1 19 1 (SD0 42 1) and this exceeds the above difference by a

Table 3 TLC estimation in 35 normal subjects bythree methods*

Chest radiograph 6-93 A 1 32 )(posture corrected) 0-72+0 37

Body plethysmograph 6-22+1-18 1 19±0-42Helium volume 574 I *20 0X48 ±0O29

(single breath) J

*Results are mean + I SD litres.

731

copyright. on M



/T


ber 1979. Dow

nloaded from



0*

r = 0-96

. .

.00@0

0*.: /z

-C0.0

C'0ELAc-

a-

2 3 4 5Plethysmograph

26 7 9

r =0 97

0

* @4

2 3 4 5 6Helium vol (10 sec) l

Fig 8 Correlation between radiographic (posturecorrected) and plethysmographic estimates of TLC in35 normal subjects. Line is line of identity.

8-

a)0.Q20l

0

2

r=0 95

.0

@0

0.0

2 3 4 5 6 7 8 9Helium volume (lOsec )

Fig 9 Correlation between radiographic and singlebreath helium estimates of TLC in 35 normal subjects.

volume of 0-48 1 (SD 0-29 1) which we attribute tothe volume of gas in the lungs that cannot bereached by a single breath of helium in 10 seconds.This can be thought of as the effective trappedgas volume. The coefficient of intraobservervariation was 0-56% (SD of the differences was

37-8 ml). Interobserver variability among the threeseparate observers gave a coefficient of variationof 4-9% (SD of the differences was 0-32 1).

Fig 10 Correlation between plethysmographic andsingle breath helium estimates of TLC in 35 normalsubjects.

Discussion

This radiographic method is both rapid and simpleto use. Estimation of lung volume from a pair ofradiographs takes less than two minutes. The use

of computers to reduce the time and tedium ofthe older manual methods has also been describedby others (Paul et al, 1974; Glenn and Greene,1975; Barrett et al, 1976). A scanning device andmicroprocessor are being developed to read thefilms and do the relatively simple sums, obviatingthe need for a computer. This will permit wideruse of the method. The main advantage of thismethod is that it uses geometric cross-sectionalshape information derived from the chests ofliving subjects to arrive at a completely inde-pendent measurement of the displacement volumeof the lungs. The estimates of lung volume that itproduces correlate closely with those of the otherindependent methods used when the differencesbetween what each method measures, and posturaleffects, are taken into account.

Magnification factors can be simply andaccurately measured even under non-standardradiographic conditions. Although it is possible touse individual factors for each of the thoracicstructures, the errors introduced by assuming thatthe heart and spine lie at the true centre of thechest in both PA and lateral projections and usinga single magnification factor applicable to thiscentral plane are small, and are minimised by

732

8-

7-

m 6-0

015-

-D0

A- 4--C

3-1

27 8 9

-w a I I

.

.

.1.

copyright. on M



/T


ber 1979. Dow

nloaded from



using a long target-film distance that reduces allfactors towards unity. Minimising the distance be-tween the patient and the x-ray film is alsoimportant in this regard.

Close co-operation between the radiographerand the subject in ensuring that the subject in-spires to TLC is essential, and the advantage ofletting the subject signal the radiographer toexpose the radiograph when he has inspired fullymay be of considerable importance in patientswith disease of the airways who have a longinspiratory time.A mean posture correction was applied to the

radiological TLC estimates of all subjects in thisstudy for comparison with the estimates by theother methods. This was done because the timeand effort spent in measuring postural effectsfor every patient on a routine basis would beconsiderable.The mean value for the difference between

radiographic and plethysmographic volumes in the35 normal subjects was +0-72 1 (SD 0-37 1). Thisdifference should be accountable for by pulmonaryblood and tissue volumes, and in fact is veryclose to the combined values for these volumesmeasured by other methods. The mean pulmonaryblood volume for subjects of the height and weightof those in this study measured by the dyedilution method of McGaff et al (1963) is about420 ml. Their tissue volume by the method ofSackner et al (1975) using an acetylene rebreath-ing technique is about 280 ml. The combination ofthese other estimates gives a volume of blood andtissue of about 700 ml for our subjects. The agree-ment with our estimate is quite close, although thedata of Peterson et al (1978) and Petrini et al(1978) suggest that the acetylene method mayslightly overestimate tissue volume in normallungs.The standard deviation of the difference be-

tween radiographic and plethysmographic volumeswas rather large and the problem with estimationof lung tissue and blood volume in this way isthat of obtaining a relatively small number fromthe difference between two large ones, each ofwhich has its own appreciable error. For thisreason, although the mean value for the wholegroup is a reasonable one, less reliance can beplaced on that for any individual subject. In thisregard it should be noted that the soluble gastechniques of estimating lung water are good atestimating small volumes of quickly accessiblefluid but greatly underestimate larger, more slowlyavailable pools of water (Denison et al in prepara-tion). Thus the two techniques are strictly

733

complementary.The mean difference between plethysmographic

and single breath helium dilution volume in thisgroup of normal subjects was +0-48 1 (SD 0-29 1),and that between radiographic and helium dilutionvolumes was + 1 19 1 (SD 0-42 1). These differencesmay provide a useful index of inhomogeneity ofventilation or gas "trapping" in subjects withairways obstruction, and the combination ofradiographic and gas dilution techniques permitssuch measurements to be made without a bodyplethysmograph.The assessment of cross-sectional shape of the

heart from CT scans in this study was complicatedby the fact that since the scanner did not havethe facility for gating with respect to the cardiaccycle, the cardiac image seen on the scans is acomposite of the movements of the heart withinthe chest as well as its size and shape. Thisaccounts for the rather large mean cardiac volume(101 1) obtained from the CT scans. In the ana-tomical sections, however, shape of the heart waseven more accurately approximated by the ellipti-cal estimate and the mean volume was a morerealistic 0'77 1. In practice the chance that theheart should be in exactly the same position andstate of contraction during the two instants atwhich the PA and lateral radiographs are taken issmall. Mean radiological volume for the heart inthe 35 normal subjects was 0-80 1 (SD 04145 1).Since this volume is relatively small comparedwith those of the whole chest and lungs, smallerrors due to slightly inaccurate shape assessmentare unimportant in lung volume estimation. Themethod can be used to measure cardiac volumefrom radiographs and thus may be of use in thestudy of patients with cardiac disease.

Since the radiographic method is extremelyaccurate for structures of known geometric shape,its accuracy for estimating lung volume is depen-dent on that with which the actual volumes ofeach of the thoracic structures can be estimatedfrom the appropriate PA and lateral dimensions.Combination of the error standard deviations forthe chest, heart, spine, and subdiaphragmaticvolumes yields an overall accuracy for themeasurement of lung volume of +fi210 ml. This iscomparable with the accuracy of the bodyplethysmograph in the estimation of TLC whichhas been assessed by Barrett et al (1976) as +250ml, consisting of components of +170 ml each forthe thoracic gas volume at FRC (DuBois et al,1956) and the inspiratory capacity components ofthe plethysmographic method. The reproducibilityof the radiographic method is dependent on that

copyright. on M



/T


ber 1979. Dow

nloaded from



of a pair of radiographs taken at TLC and thatwith which a given set of radiographs can bedigitised. The former is difficult to assess becauseof the radiation exposure required for multiplepairs of radiographs in the same subject. Thelatter is <1% for a trained observer.

We thank Dr I Kerr and the staff of the RadiologyDepartment of the Brompton Hospital and Mr CBennals of the Radiology Department of KingEdward VII Hospital, Midhurst, for help andadvice throughout the study. RJP was supportedby a generous grant from the Asthma ResearchCouncil.

References

Barnhard, H J, Pierce, J A, Joyce, J W, and Bates,J H (1960). Roentgen9graphic determination ofTLC. A new method e4aluated in health, emphy-sema and congestive heat failure. American Journalof Medicine, 28, 51-60.

Barrett, W A, Clayton, P D, Lambson, C R, andMorris, A H (1976). Computerized roentgenographicdetermination of total lung capacity. American Re-view of Respiratory Disease, 113, 239-244.

Brown, R, Hoppin, F G, jun, Igram, R H, jun,Saunders, N A, and McFadden, E R, jun (1978a).Influence of abdominal gas on the Boyle's law deter-mination of thoracic gas volume. Journal of AppliedPhysiology: Respiratory, Environmental and Exer-cise Physiology, 44, 469-473.

Brown, R, Ingram, R H, and McFadden, E R, jun(1978b). Problems in plethysmographic assessmentof total lung capacity in asthma. American Reviewof Respiratory Disease, 118, 685-692.

Dubois, A B, Botelho, S Y, Bedell, G N, Marshall, R,and Comroe, J H (1956). A rapid plethysmographicmethod for measuring thoracic gas volume: A com-parison with a nitrogen washout method formeasuring functional residual capacity in normalsubjects. Journal of Clinical Investigation, 35, 322-326.

Eycleschymer, A C, and Schoemaker, D M (1970). A

Cross-section Anatomy. Butterworths, London.Glenn, W V, jun, and Greene, R (1975). Rapid

computer-aided radiographic calculation of totallung capacity (TLC). Radiology, 117, 269-273.

Habib, M P, and Engel, L A (1978). Influence of thepanting technique on the measurement of thoracicgas volume. American Review of RespiratoryDisease, 117, 265-271.

Hurtado, A, and Fray, W H (1933). Studies of totalpulmonary capacity and its subdivisions. II Cor-relation with physical and radiological measure-ments. Journal of Clinical Investigation, 12, 807-823.

Kovach, J C, Avedian, V, Morales, G, and Poulos, P(1956). Lung compartment determination. Journalof Thoracic Surgery, 31, 452-457.

Matsukawa, A, Ito, T, and Kimura, K (1977). Cross-section Anatomy and Computed Tomography:Trunk. Igakutosho Shuppan, Tokyo.

McGaff, C J, Roveti, G C, Glassman, E, and Milnor,W R (1963). The pulmonary blood volume in rheu-matic heart disease and its alteration by isopro-terenol. Circulation, 27, 77-84.

Paul, J L, Levine, M D, Fraser, R G, and Laszlo, C A(1974). The measurement of total lung capacitybased on a computer analysis of anterior and lateralradiographic chest images. IEEE Transactions onBiomedical Engineering, 21, 444 451.

Peterson, B T, Petrini, M F, Hyde, R W, andSchreiner, B F (1978). Pulmonary tissue volume indogs during pulmonary edema. Journal of AppliedPhysiology, 44, 782-794.

Petrini, M F, Peterson, B T, and Hyde, R W (1978).Lung tissue volume and blood flow by rebreathing:Theory. Journal of Applied Physiology, 44, 795-802.

Sackner, M A, Greeneltch, D, Heiman, M S, Epstein,S, and Atkins, N (1975). Diffusing capacity, mem-brane diffusing capacity, capillary blood volume,pulmonary tissue volume and cardiac outputmeasured by a rebreathing technique. A mericanReview of Respiratory Disease, 111, 157-165.

Requests for reprints to: Dr D M Denison, Lung Func-tion Unit, Brompton Hospital, Fulham Road, LondonSW3 6HP.

734

copyright. on M



/T


ber 1979. Dow

nloaded from


estimation lung · and helium-dilution estimates of total lung capacity (tlc), taking account of...

Documents