The Value of Portable Chest Roentgenography in Adult Respiratory Distress Syndrome

The Value of Portable Chest Roentgenography in Adult Respiratory Distress Syndrome* Comparison

with Computed

Tomography

Michela Bambino, M.D.;t Luciano Gattinoni, M.D.;:J; Antonio Pesenti, M.D.;§ Massimo Pistolesi, M.D.;~ and Massimo Miniati, M.D.II

In 17 patients with adult respiratory distress syndrome, we used data derived from computed tomographic (CT) scan densitometric analysis to validate the value of portable chest roentgenograms in objectively estimating the amount of pulmonary edema. Chest roentgenograms and CT scans were taken in the same ventilatory coaditioos (apnea at 10 em H.O of positive end-expiratory pressure [PEEP]); blood gas samples and hemodynamic parameters were collected at the same time. Roentgenographic analysis was undertaken by independent observen using two standardized scoring systems proposed in the literature. CT scan analysis was perfOrmed using the CT number frequency distribution and the gas lung volume (measured by helium dilution technique) to estimate quantitatively the lung density, the lung weight, and the percentage of normally aerated and nooaerated tissue. Knowing the mean CT number of the pulmonary parenchyma in a group of normal subjects, we also inferred the ideal lung weight expected in the study population and computed the excess tissue mass as the difference between actual and ideal lung weight. Both the

roentgenographic scoring systems showed direct correlation with the pulmonary impairment as detected by CT scan densitometric analysis (CT number, percentage of oooaerated tissue, lung weight, and excess tissue mass; p<0.01) and inverse relation with the percentage of normally aerated tissue (p<0.01). We also found a relationship between roentgenographic scores and the impairment in gas exchange as detected by shunt fraction (p
The presence of diffuse alveolar infiltrates on the chest roentgenogram is a characteristic feature of the adult respiratory distress syndrome (ARDS), a clinical disorder consisting of parenchymal inflammation, permeability pulmonary edema, severe dyspnea, and refractory hypoxemia.•-3 The accumulation of fluid in the extravascular space, by replacing the gas density with a water-equivalent density, yields a progressive increase in the overall lung tissue density that is visualized on the chest roentgenogram.

The best approach to evaluate lung edema in patients with ARDS should be the direct measurement of the amount of extravascular lung water (EVLW). However, the assessment of EVLW by indicator dilution techniques is limited by the invasiveness of the procedures and by the possible regional abnormalities in lung perfusion, leading to an underestimate of EVLW. 4 •11 Vascular obstructions have been, indeed, extensively reported in autopsy lung specimens from ARDS patients6 and also have been demonstrated clinically by bedside pulmonary angiography7 •8 and perfusion lung scanning. 9 More sophisticated techniques for lung water measurement, such as nuclear resonance 10 or positron emission tomography11.1 2 are restricted in their application by high cost and low practicality. Hence, at the present time, the chest roentgenogram is the only practical means available to evaluate pulmonary edema in critically ill patients. 13 Chest roentgenography provides excellent information about regional distribution of lung edema and, unlike indicator dilution techniques, does not depend on active

*From the Institute of Anesthesia and Intensive Care, University of Milan, Milan, Italy. tCiini<.-al Assistant in Anesthesia and Intensive Care, San Gerardo Hospital, Monza. :!:Professor in Anesthesia and Intensive Care. §Research Fellow, Institute of Anesthesia and Intensive Care. 'Associate Professor of Respiratory Pathophysiology II, University of Rome. !!Career Investigator, CNR, Institute of Clinical Physiology, University of Pisa. This work was supported in part by grants 87.00750.57 and 87.00018.26.115 from C.N.R., Rome. Manuscript received September 13; revision accepted February 13. &print requests: Dr. BOmbtno, 1 Servlzio Anute.ria e IUarHmclzione, Ospedale NUQfXJ S. Gerardo, Monr.a (MI), Italy 20052

782

ARDS =adult respiratory distress syndrome; ABF =acute respiratory failure; CI =cardiac index; CO= cardiac output; CI=computed tomography; EVLW=extravascular lung water; FRC = £unctiooal residual capacity; ICU =intensive care unit; PAP=pulmonary artery pressure; PEEP=positive endexpiratory pressure; Qsp/Qt = right-to-left shunt fraction; TSLC =compliance of the total respiratory system; WP = pulmonary capillary pressure

Port1ib1e Cheat RoelllgiiiOgl'llllhy In AROS (Bomb/no et el)

arterial perfusion of edematous areas. Evaluation of lung density by chest roentgenography is, however, affected, to a various extent, by the degree of lung inflation, 14 the superimposition of lung structures with different density, and the quality of the roentgenography 6lm, especially when obtained under emergency conditions. Furthermore, it is widely objected that the evaluation of roentgenographic findings may be affected by the subjective interpretation of the reader. With the present study, we aim to reappraise chest roentgenography as a valuable method to assess lung edema in patients with acute respiratory failure (ARF) by comparing a standardized reading of portable chest roentgenograms with the determination of lung tissue density obtained by computed tomography (CI'). 111 MATERIAL AND METHODS

lbtfent lbpulafion

This study was conducted in the frame of a project for the study of ARDS approved by the Human Studies Committee of the Lombardia Region. IDformed ooosent was obtained from the patients or relatives before the study.

The study group consisted of 17 patients (11 males, six females; mean age, 41 ± 18 years) with ARDS of various causes (four bacterial pneumonia; llve viral pneumonia; four polytrauma; and four otben (peritonitis, sepsis, pulmonary embolism, encepbalitisD; there were 11 survivors and six nonsurvivors. All the patients were intubated and mechanically ventilated and had a ftow.directed thermodilution Swan-Canz catheter and an arterial cannula inserted, prior to the study, tOr clinical reasons. No patient had a history of lung disease. ~lnlcedunr

The patients were studied after a mean of 4±3 days (range, 1to 11 days) after the onset of full-blown respiratory failure as indicated by the intubation time. All the patients were kept in pualysis (pancuronium bromide) and intravenous anesthesia (Fentanyl, thiopental sodium) was given throughout the study. The chest roentgenograms were taken in the intensive care unit

Table

(ICU) during apnea with the lung inllated at 10 em H 10 positive in the supine position. Portable end-espiratory pressure (PEEP~ chest roentgenography equipment was used (Toshiba); the electron power ranged between 48 and 156 kV according to the patients size; the exposure time was kept below 0.2 s, while the distance between the focus and the lllm plane was 120 em. Within 1 to 2 h of the chest roentgenograms, the patients were transferred to the Radiology Department to undergo the cr scan (Pfizer AS/EO Cf scanner). The exposures were taken at 120 kV. 50 mA, and 5 s. Slice thiclmess was 9 mm and the dimensions of the pixels of the reconstruction matrix were 1.5 X 1.5 mm x 9 mm. The system was previously calibrated by a suitable phantom. scan, the mechanical ventilation was delivered at Before the the same parameters used in ICU before the chest roentgenogram (fe, Flo., 0.6 to 1.0, according to clinical conditions; tidal volume, 10 mllrg-•; respiratory rate, 16to 22 beats per minute; square wave form; inspiratory-expry ratio, 1:1; inspiratory pause, 0; PEEP, 10emH10). The llrst level of exposure was above the carina (apex~ the second was approximately at the hilum (hilum), and the third was above the diaphragm (base). Every exposure was performed with the patient in apnea at 10 em H.O PEEP, as during chest roentgenograms.

cr

Plapologfc lbrameten The functional residual capacity (FRC) was measured at atmospheric pressure by helium dilution technique, while the compliance of the total respiratory system (TSLC) was measured by a supersy· riDge through step-IJr-step inflation-detlation (100 ml step) and the volume pressure curve was then drawn manually.,. The lung volume at 10 em H.O PEEP was computed adding to the measured FRC the volume computed on the in8ation limb of the VIP curve to reach 10 em H.O pressure. Blood samples, arterial (arterial cannula) and mixed venous (Swan-Ganz catheter~ were taken after the scan and analyzed tOr ps tensions (Po., Pea.. IL 1312 pH/Blood Gas Analyzer) and oxygen saturation (IL CO-Oximeter 283~ The right-to-left shunt fraction (QspiQt) was computed according to the standard formula. The mean pulmonary artery pressure (PAP) and the mean pulmonary capillary pressure (WP) were measured by transducen (Bentley Trantec) and recorded by monitor (128 Kontron); cardiac output (CO) was measured in triplicate by thermodilution (COEdwards) and cardiac index (CI) was computed.

cr

Scoring tfl.tmg EderntJ in ABF

1-~

Score Bt

Score A•

X-Ray Finding Right-sided cardiac enlargement with bulging of main pulmonary artery Hilar abnormalities (size and density) Air bronchogram Lung density Increase Hazy

Central Peripheral

Central and peripheral Patchy Central Peripheral Central and peripheral Extensive white density

X-Ray Finding

Score

2,4 1,2 2,4

1,2 2,4 3,6 2,4 5,10 7,14 20

Normal Pulmonary vascular congestion Mild

Moderate

Severe Interstitial edema without septal lines Interstitial edema with septal lines Mixed interstitial and alveolar edema With some sparing of pulmonary region Involving entire region

Score

0 10 20

30 40

45 50 155

Alveolar edema With some sparing of pulmonary region Involving entire region

60 65

•Each hmg is scored separately: score range, 0 to 52 (Pistolesi~ tThe right and left lungs were divided into a total of six regions: two upper, two lower, and two perihilar regions: score range, 0 to 390 (Halperin). CHEST I 100 I 3 I SEPTEMBER, 1881

713

Chelf Roentgenography Analym

Table !-Image AnGlpia

The chest roentgenograms were analyzed in another institution by two trained independent observers (M.P., M.M.) who had no infonnation about the clinical course, physiologic parameters, and cr scan data. To standardize the analysis, we used two reading tables that have been described in previously published reports.•...• The two scoring systems (A and B) are reported in 'lllble 1; both consider the severity, extension, and regional distribution of roentgenographic signs of lung edema. In addition, score A also takes into account cardiac and hilar abnormalities that were found to have a high prevalence in ARDS patients.'" Since these latter roentgenographic findings cannot be considered, per se, specific for lung edema, they were assigned a low numeric value. In score A, the features of the roentgenographic increase in lung density considered for analysis were chosen to correspond to three difFerent stages of lung involvement, from pure interstitial edema ("hazy," fe, ill-defined densities that allow the underlying vascular structures to be seen) to alveolar edema \'patchy;· fe, weD-defined densities obscuring the underlying vascular structures). and to extensive lung consolidation (referred as to "extensive white density"). Peripheral distribution of the increased lung density was given a higher score than central distribution since it represents a more severe degree of lung involvement.•• Intraobserver variability was assessed by computing the average variation around the mean value of two successive scores assigned by each reader two weeks apart. Ukewise, interobserver variability for both scoring procedures was estimated by comparing the mean value of the two independent readers. The average roentgenographic score (mean offour scores assigned to each roentgenography by the two observers) was used to compare the two scoring systems with each other and with cr scan data and physiologic parameters.

Meaner No. NonnaUy aerated tissue, ._, Poorly aerated tissue, ._, Nonaerated tissue,._, Mean lung weight, g kg-• Score A (scaled 0-52) Score B (scaled 0-390)

Normal Value (n=7) Mean±SD

Patient Population (n=17) Mean±SD

-699±11 77±3 16±2 7±1 19±2 Mean±SD

-379±133 20±11 26±9 54±15 40±19 27±12

Range

1.2-48

Mean±SD Range

9.18±79 130-365

that the cr number reftects the physical density, fe, the ratio between the volume of gas plus the volume of gas plus the volume of tissue, and being the total amount of gas present in the lung known by an independent measurement (FRC by helium dilution). we could estimate the lung tissue weight (considering the specific weight of the lung tissue = 1) rearranging the following equation: Volume of Gas VolumeofGas +Tissue Mean cr Number ofthe Whole Lung CfNumberofAir- CfNumberofWater Moreover, knowing the cr frequency distribution, the weight for each compartment (nonnaUy aerated, poorly aerated, and nonaerated) was computed. The lung weight, computed as above, includes the lung structural mass, the blood, and the EVLW. The nonnal expected values (ideal lung weight) for a given patient were computed according to the above equation by substituting as gas volume the espected normal FRC (according to the equation proposed by Ibanez and Rauric"" for supine and paralyzed patients). and substituting as mean cr number the mean cr value we found in normal subjects at atmospheric pressure. (It is worth emphasizing that the dispersion of cr number, in nonnal individuals, is very narrow. See 'liable 2). The percentage of increase in lung tissue mass (excess tissue mass) was then computed as the dift'erence between actual and espected normal lung weight indexed to the latest.

CT Scan Anolym The method we used for cr scan analysis has been described elsewhere"" and will be only summarized herein. BrieRy, the single transverse areas of the thorax were measured planimetricaUy, drawing manuaUy the boundary of each lung along the Inside of the ribs and the edge of the mediastinal organs. The surface of the contoured areas was computed (PDP 11.123 Digital Computer, liT II operative system). The total sectional area of the lung was then obtained as the sum of each of the six areas considered (two apical, two hilar, and two basal). The quantitative analysis of cr scan image was perfonned by a computer program designed to provide the mean cr number of the total cross-sectional area and the cr number frequency distribution. Pixels showing a cr number between -1,000 Hand -500 H were considered "nonnal" (nonnaUy aerated tissue). between -500 H and -100 H "medium dense" (poorly aerated tissue), and between -100 Hand + 100 H "abnonnally dense" (nonaerated tissue). This subdivision is founded on data from a larger population of ARDS patients and the comparison with the cr number distribution observed In nonnallungs. ••

StlltVtical Anolym

The results are reported as mean± standard deviation. Coefficients of linear correlation and regression lines were calculated according to standard methods. REsuLTS

Physiologic and Imaging lbrameters of the Study Population Some of the most relevant gas exchange, lung mechanics and hemodynamic parameters of the patient population at the time of the study are summarized in Thble 3. These parameters were taken with the patient in paralysis and anesthesia during mechanical ventilation with PEEP of 10 em H20. th The gas exchange parameters are consistent wi

~ht ComputatWn W! assumed that the three cr scan sections obtained (apex, hilum, and basis), fe, the total cross-sectional lung surface area, are a representative sample of the whole lung and we considered the lung as made of a mixture of water and air only. W! then obtained an estimate of the mean cr number of the whole lung. Knowing I..ung

Table 3-Piay8iologic lbrarnelerl

Mean+SD

784

Qsp!Qt

PaO,, mmHg

Lmin·•

co.

PAP, mmHg

mlcm u.o-•

FRC, ml

0.40+0.15

71+19

8+3

32±4

44±19

1050+463

TSLC,

Portable 018111 RollltgiiiOglllllhy In ARDS (Bomblno et al)

FIGURE I. Portable chest roentgenogram and Cf scan (apex, hilum , a nd basis) appearance in a patient with ARDS due to peritonitis. Roentgenographic scores were 44 (range. 12 to 48) for score A and 349 (range, 130 to 365) for score B. Cf scan analysis showed a percent of nonaerated tissue of 65.6 percent . while the normally aerated tissue was ll percent. The mean lung weight for body weight was 56.76 g, four times the normal value. The shunt fraction in this patient was 44 perl't'nt.

moderate to severe ARDS, according to previously published criteria.23 •24 The quantitative data derived from CT scan analysis are reported, in average , in Table 2, as well as the mean score obtained by the two roentgenographic scoring procedures (A and B). Reference values obtained in seven normal subjects are added for comparison. Figure 1 shows a typical ARDS imaging study, as detectable by portable chest roentgenogram and CT scan, with the patient in the supine position and the 400 350

•

a>

w

300

u

250

0:: 0

(Jl

::E:

-"'

200

a.: <.:>

150

a:

100

0

0::

y-71 i'6 . 07x r•0.97;p <0.001

50 0

Chest Roentgenography Scoring System The intraobserver variability averaged 6.1 percent (range, 0 to 22 percent) and 6.7 percent (range, 0 to 19 percent), considering score A, and 5.8 percent (range, 0 to 23 percent) and 9.6 percent (range, 1 to 31 percent) considering score B. The interobserver variability averaged 9.2 percent (range, 0 to 28 percent) and 6.1 percent (range , 0 to 17 percent) for score A and B, respectively. As shown in Figure 2, the two scoring systems were significantly correlated in linear fashion over the entire range of score assigned .

Relationship between Chest Roentgenography and CT Scan Analysis

u

a. a:

same degree of lung inflation (during apnea, 10 em H 20 PEEP).

CSI

"'

~

~

RADIOGRAPHIC SCORE A FIGURE 2. Relationship between the two roentgenographic scoring systems (see Table 1).

All the variables derived by the quantitative CT scan analysis were significantly associated with the standardized roentgenographic readings. Figure 3 shows the relationship we found hetween CT scan density and the scores A and B. The roentgenographic scores A and B were also directly related to the fraction of nonaerated tissue and inversely related to the fraction of normally aerated tissue, as shown in Figures 4 and 5, respectively. Finally, both the lung weight (expressed in g kg - •) CHEST I 100 I 3 I SEPTEMBER , 1991

765

50 45

a: 40

~

•

35

i.

u 30

•• •

4BB,---------------------~ 350

•

•••

... 20

il5

lB

lBB

y•S I. 7+8. 964lx r•S.7GI;p
y•393. S+GI. 4lx r•lil.73;p
50

0

0

'~' ' '~m

~-'-....._.

....... ...~-....._.......,.J-.......__._.J-...J

'~' ' ' m

mean CT value (H units) mean CT value CH units) FlcURB 3. Relatiooahlp between the mean cr number of the whole I1JD11111d the IClOI'iDg l)'lteml A (left panel) and B (right panel).

and the excess tissue mass were strictly relMed to the scores A and B according to the followiDg equations: Mean Lung Weight==9.72+1.12 Score A (r==0.71, p
(roentgenographic scores) data were related to the severity of ARDS evaluated by the impairment in gas exchange. The nonaerated tissue, as detected by CT scan, and both A and B scoring values were in fact significantly correlated with the shunt &action (Qspl Qt): Qsp/Qt vs percent nonaerated tissuer==0.61, p<0.05 QspiQt vs score A r==0.61, p<0.05 Qsp/Qt vs score B r==0.53, p<0.05

Imaging and Physiologic EbrCJf'f18fer8

During the last few years, we used data derived from CT scan analysis to investigate the mechanisms

DISCUSSION

...

Both the parametric (CT scan) and nonparametric

•

50 45

a:

40

~

35

u

3B

Ill

~

i. i

...

31118 250

u

iI...

2B IS 10

•

350

y•-4.2+GI.S8x r•GI.7l;p
NON AERATED TISSUE C% of lung •eight)

2lill

l5B lee

y•49.7+3.47x r•G1.69;p<8.8l

50

NON AERATED TISSUE (~

of lung

~lght)

FICURB 4. Relationahip between the percent of noaaerated tissue lllld the two IClOI'iDg systeml A (left panel) lllld B (right panel).

50 45

a: Lo.l

IX 0

u

en u.....

if a: IX C)

0 .....

Cl

a: 0::

40

400

•• •

• •

350 IZI

w

300

IX

35

0

u

en u.....

30

::z:: ~ a:

25

250

200

•

IX

C)

20

•

IS

10

y•42. S-0. 7Sx r•0.70;p<0.01

5

0 .....

Cl

a: 0::

150 100

y•344. 3-4. Six

50

r•0.72;p<0.01

0L...o......r........-..L.........a._._L.......J-,.,-..L..........J...............-..L..~

m

NORMALLY AERATED TISSUE (~

of lung weight)

~

m

~

--N N

m

~

m ~

~

~

m ~

NORMALLY AERATED TISSUE (~

~

~

m ~

of lung weight)

FIGURE 5. Relationship between the percent of nonnally aerated tissue and roentgenographic scoring systems A (lef\ panel) and 8 (right panel).

by which the application of a PEEP may affect gas exchange in ARDS patients (recruitment oflung zones expressed as decreased lung densities)!S and to relate the Cf scan data to the mechanical proprieties IJf the diseased lung. 116 The population in this study is part of that larger group of ARDS patients in whom a portable chest roentgenogram in standardized condition was performed for comparison with cr scan image. The clinical value of chest CT scan as a diagnostic tool in pulmonary medicine is surely not in discussion and received considerable attention in the last years. 27-29 Routine portable chest roentgenograms still have the greatest application in the assessment oflung ~mpairment in critically ill patients, and the information derived from comparison of roentgenograms is used daily to evaluate the occurrence of new parenchymal infiltrates or pleural effusions or in emergency situations (tension pneumothorax, hyperinflation, etc); a CT scan is necessary when the clinical course diverges from the roentgenographic findings, and it is still limited in its applicability beca~se of the cost and mainly because of the cumbersome transport of a critically ill patient to the cr room. Both CT scan and chest roentgenography give us information about densities, their distribution pattern (morphology), and their entity. One of the hallmarks of ARDS definition, the roentgenographic appearance of diffuse parenchymal infiltrates, has been questioned by Cf scan studies performed by our group and others30 •31 that pointed out a prevalent localization of lung densities in the dependent zones of the lung with preservation of regions with near-normal gas/tissue ratio. However, the topography of densities seems not to affect the sensitivity of the scoring systems in detecting the overall lung impairment as shown in Figure 1 where a four times increase in lung weight

compared with normal value (as computed by Cf scan data) correlates with comparable high roentgenographic scores. No attempt was made to have a more uniform population, and the variability between physiologic parameters can be due to different timing in the evolution of the illness at the moment of the study (1 to 11 days from intubation) and accounts for the different degree of lung impairment, as assessed by Cf scan (Tables 2 and 3). The physiologic parameters in Table 3 describe the degree of functional deterioration in gas exchange, pulmonary mechanics, and hemodynamics, while the comparison with a normal population in Table 2 suggests the degree of "structural" impairment as detected by Cf scan analysis, with a two times increase in lung weight in our population when compared with the mean value in seven volunteers comparable for age and anthropometric parameters. To assess the objective value of the S(.'Oring systems proposed to evaluate the amount of pulmonary edema, we decided to keep the two readers completely unaware of the clinical history and course, as well as of physiologic and CT scan data, to avoid possible biases due to patient knowledge. The low intraobserver and interobserver variability points out that a scoring system can enhance the semiquantitative value of chest roentgenography, leaving less subjectivity in the evaluation of lung impairment. Others32 used arbitrary grades of pulmonary infiltrates to quantify lung edema on chest roentgenograms; conversely, the scoring systems we used are based also on the recognition of specific findings correlated to the presence of edema that are graded and contribute to the final score assigned to the chest roentgenogram under examination. This may enhance the ability of the CHEST I 100 I 3 I SEPTEMBER, 1991

767

reader to detect small changes in the roentgenogram, leading to less misinterpretation of the lung impairment. Both the scoring systems evaluated in this study seem to agree in the estimation of the degree of pulmonary edema (Fig 2), despite the different procedure used in the analysis of the chest roentgenograms. The use of data from cr scan image analysis for the validation of roentgenographic scores has its rationale in the objective quantification of lung impairment assessed by Cf scan density analysis. The regressions in Figures 3, 4, and 5 show a good correlation between Cf scan data and roentgenographic scores. However, we have to point out that the scatter of data points in these regression lines can reflect some of the limitations of the semiquantitative methods in chest roentgenography readings (overestimation of density in the supine position or underestimation of density in an area with relative hyperinflation, for example). On the other hand, cr scan analysis does not allow differentiation of densities due to EVLW and consolidation or blood pooling. A previous work32 comparing roentgenographic detection of pulmonary edema with the assessment of EVLW by thermal-dye dilution technique concluded that the chest roentgenogram does not give a good estimation of lung edema. The authors made the a priori statement that EVLW measurements by thermal-dye dilution gives the real assessment of the amount of pulmonary edema. To assess the actual EVLW in a given lung, the thermal-dye indicator is supposed to "see" all the areas of the parenchyma under investigation. If perfusion does not occur in some zones, as described iri ARDS lung, ~>- 9 these are excluded from analysis leading to an underestimation of the amount of pulmonary edema. Conversely, the high diffusibility of the thermal indicator may lead to a overestimation of edema due to the possibility of "reading" the blood distal to microemboli as extravascular.5 In the same article, 32 the authors did not find any correlation between "functional" deterioration, as assessed by the degree of Qsp/Qt, and the assessment of EVLW by thermal-dye dilution technique. In our study, both the roentgenographic scores and the Cfderived data were correlated to the degree of gasexchange impairment of the diseased lung. However, our patient population was more uniform than that reported in the other article, all of our patients having noncardiogenic pulmonary edema, and we cannot infer from our data that this would apply to other patients with ARF not due to ARDS. Pistolesi et al,1 7 in fact, developed different scoring procedures for cardiogenic and noncardiogenic pulmonary edema. The good correlations we found between the roentgenographic scores and both the actual lung weight 788

and the excess tissue mass computed by cr scan analysis leistolesi M, Miniati M, Di Ricco G, Marini C, Giuntini C. Perfusion lung imaging in the adult respiratory distress syndrome. J Thorac lmag 1986; 1:11-24 10 Cutillo AG, Morris AH, Ailon DC, Durney CH, Case TA: Determination oflung water content and distribution by nuclear magnetic resonance imaging. J Thorac Imaging 1986; 1:39-51 11 Shuster DP, Marklin GF, Mintun MA. Regional changes in extravascular lung water detected by positron emission tomograph}: J Appl Physioll986; 60:1170-78 12 Calandrino FS Jr, Anderson OJ, Mintun MA, Shuster PP. Pulmonary vascular permeability during the adult respiratory distress syndrome: a positron emission study. Am Rev Respir Dis 1988; 138:421-28 13 Staub NC. Clink-al use of lung water measurements: report of a workshop. Chest 1986; 90:588-94 14 Zimmerman JE, Goodman LR, Shahvari MGB. Effect of mechanical ventilation and positive end-expiratory pressure (PEEP) on chest radiograph. AJR 1979; 133:811-15 Portable Chest Roet llgel1011f81111Y in ARDS (Bombino et al)

15 Hedlund L\V, '\bck P, Ellinan EL. Computed tomography of the lung. Radiol Clin North Am 1983; 21:715-88 16 Gattinoni L, Mascheroni D, Basilica E, Foti G, Pesenti A, Avalli L. \Ulumelpressure curve of total respiratory system in paralyzed patients: artifacts and correction factors. Intensive Care Med 1987; 13:19-25 17 Pistolesi M, Miniati M, Milne ENC, Giuntini C. The chest roentgenogram in pulmonary edema. Clin Chest Med 1985; 6:315-44 18 Halperin BD, Feeley Tw, Mihm FG, Chiles C, Guthaner DF, Blank NE. Evaluation of portable chest roentgenogram fur quantitating extravascular lung water in critically ill adults. Chest 1985; 88:649-52 IS Milne ENC, Pistolesi M, Miniati M, Giuntini C. The radiologic distinction of cardiogenic and noncardiogenic edema. AJR 1985; 144:879-94 20 Gattinoni L, Pesenti A, Torresin A, Baglioni S, Rivolta M, Rossi F, et al. Adult respiratory distress syndrome pro&les by computed tomography. J Thone Imaging 1986; 1:25-30 21 Rosenblum LJ, Maurcerl RA, WeUenstein DE, Thomas FD, Bassano DA, Baasch BN, et al. Density pattern in the normal lung as determined by computed tomography. Radiology 1980; 137:409-16 22 Ibanez J, Rauric JM. Normal values of functional residual capacity in the sitting and supine position. Intensive Care Med 1982; 8:173-77 23 Murray JF, Mattru.y t.1A, Luce JM, Flick MR. An expanded de&nition of the adult respiratory distress syndrome. Am Rev Respir Dis 1988; 138:720-23

24 Fowler AA, Hamman RF, Zerbe GO, Benson KM, Hyres TM. Adult respiratory distress syndrome: prognosis after onset. Am Rev llespir Dis 1985; 132:472-78 25 Gattinoni L, Pesenti A, Bombino M, Baglioni S, Rivolta M, Rossi F, et al. Relationships between lung computed tomographic density, gas exchange, and PEEP in acute respiratory failure. Anesthesiology 1988; 69:824-32 26 Gattinoni L, Pesenti A, Avalli L, Rossi F, Bambino M. \Uiumepressure curve of total respiratory system in acute respiratory failure: CT scan study. Am Rev Respir Dis 1987; 136:730-36 27 Peruzzi W. Garner W. Bools J, Rasanen J, Mueller CF, Reilley T. Portable chest roentgenography and computed tomography in critically ill patients. Chest 1988; 93:722-26 28 Golding RP, Knape P, Strack Van Schijndel RJM, DeJong D, Thijs LG. Computed tomography as an adjunct to chest X-rays of intensive care unit patients. Crit Care Med 1988; 211-16 29 TocinQ I, Miller MH. Computed tomography in blunt chest trauma. J Thone Imaging 1987; 2:A 30 Maunder RJ, Shuman WP, McHugh }w. Marglin SI, Butler J. Preservation of normal lung regions in the adult respiratory distress syndrome: analysis by computed tomography. JAMA 1986; 255:2463-65 31 Gattinoni L, Pesenti A. ARDS: the non-hc,mogeneous lung; facts and hypothesis. Intensive Crit eire Digest 1987; 6:1-4 32 Sivak ED, Richmond BJ, O'Donovan PB, Borkowski GP. Value of extravascular lung water measurement vs portable chest X-ray in the management of pulmonary edema. Crit Care Med 1983; 11:498-501

CHEST I 100 I 3 I SEPTEMBER, 1981

781