Computed tomography and pulmonary ... - Semantic Scholar

3 downloads 60 Views 195KB Size Report
Computed tomography (CT) was first developed to measure accurately ... CT measures the radiographic density of an array of small ... resolution of the system, so that a single voxel may .... subjects, their measurements of density increased at.
British Journal of Anaesthesia 1998; 80: 665–671

REVIEW ARTICLE

Computed tomography and pulmonary measurements G. B. DRUMMOND

Computed tomography (CT) was first developed to measure accurately radiographic density, rather than as a method of imaging.19 Although subsequent development concentrated on the imaging potential of the method, some recent studies have used the ability of the method to quantify radiographic density, to provide precise measurements of lung regions and infer differences in regional properties. CT measures the radiographic density of an array of small discrete parts of the body. Usually, each of these parts in a single tomographic slice is represented as a pixel, and these pixels are displayed as the units of a two-dimensional image. The operator allocates a range of grey shades between black and white (tone range) to a particular range of radiographic densities, so that the densities of interest lie between black and white in the visual image. For example, when the details of lung tissue are of interest in the image, the density window (measured in Hounsfield units, HU) often chosen is from 9500 (black) to 9200 (white). These settings can be expressed in terms of the window centre and width. Lung tissue, containing a large proportion of air, has a range of radiographic densities much less than other body tissues, for example the soft tissues of the mediastinum. Consequently, details of lung structure can be shown more clearly by expressing this narrow range of low densities over the whole visual range from black to white. The grey scale that shows how the black and white visual image is related to radiographic density is displayed at the edge of most CT images, and can be altered by the operator to enhance the features of interest. Each CT image therefore consists of an array of small squares, each with a discrete grey value, that represents the mean radiographic density of a cylinder of tissue, or voxel. The end is represented in position and area by the pixel, and the length is the thickness of the CT slice. Although the visual image is often altered to improve the contrast of specific features, the original radiographic density values can be used to obtain more specific information about the properties of the materials studied. Radiographic density is the property of materials to absorb electromagnetic radiation. This radiation is exponentially reduced on passage through materials, so that the intensity (I) after passing a distance (x)

Partial volume effects When CT is used to measure lung density, dense tissues adjacent to the lung can influence the average slice density. This is because of the limited spatial

(Br. J. Anaesth. 1998; 80: 665–671) Keywords: measurement techniques, graphy; ventilation, regional; lung, volume

through a material can be expressed: I:Io e9␮x where I0:radiation intensity incident on the material and ␮:linear attenuation coefficient for the material. Attenuation of electromagnetic radiation is mainly by two processes, photoelectric absorption and Compton scattering. Consequently, factors that influence attenuation include the physical density of the material, and the atomic number, electron density and mass electron density of the materials. With an energy spectrum of 120–140 keV, which is the energy used for medical radiology, attenuation is mainly by Compton scattering, which depends on the electron density of the material. As the relationship between electron density and physical density is relatively constant for most materials, there is a good relationship between attenuation and electron density and a reasonable relationship with physical density. A series of exact shapes made from different woods have a linear relationship between physical and radiographic density, in the range 0.37–0.9 g cm93, but the relationship is non-linear at greater densities.34 However, water, which is an important body constituent, does not follow exactly the same relationship as other materials.34 44 Absorption of radiation which has lower energy is by photoelectric absorption, and this depends more on the atomic number of the material. Consequently, physical density and radiographic density may be variably related according to the range of electron energies from the source, because the absorption of radiation with different energies varies with the material studied.37 This is made more complex by the more marked attenuation of the lower energies (beam hardening). Variations in operating energy can cause significant error in measurements,56 and change in position of the subject in the x-ray beam can also affect the results.26 Another potential source of error is variation in detector gain, which is more likely with measurements of low attenuation materials such as the lung.56 Density assessments between different scanners may differ significantly.26 However, despite these sources of variation, repeat measurements in optimal conditions can be highly reproducible.56

computed

tomoG. B. DRUMMOND, MA, FRCA, University Department of Anaesthetics, Royal Infirmary, Edinburgh EH3 9YW.

666 resolution of the system, so that a single voxel may contain both lung tissue and also some adjacent dense tissue in the slice. This is the partial volume effect. This effect becomes greater when the overall area of the lung slice is small. However, the effect is predictable and a correction can be applied. Such slices are obviously present at the lung apex, but there are also small areas in the crescentic slices posterior to the diaphragm at the lung base.4 Pulsation of blood vessels, over the time necessary for a scan, is another factor that increases the density of a voxel. Radiographic density is most often expressed in Hounsfield units, a dimensionless index of attenuation that is related to the attenuation of air and water. The attenuation coefficient of air is allocated a value of 91000, and water is allocated an attenuation coefficient of zero. The attenuation value, CTm:1000. (␮m– ␮w)/(␮w– ␮a) where ␮w, ␮a and ␮m:attenuation values for water, air and the tissue being measured, respectively. If a linear relationship is assumed between Hounsfield number and density, then radiographic density values from CT provide a non-invasive and anatomically localized index of tissue density. In the lung, the relative content of air and solid material (either tissue, or blood and other fluid such as oedema) can be calculated with reasonable confidence. The volume of solid organs can be measured to within 5% using CT.1 Determination of the boundaries between lung and chest wall can be difficult, particularly if collapsed lung tissue is present.20 Studies have used threshold values of 9200 HU to distinguish lung from soft tissue, although in some disease states lung tissue may have densities greater than this. Operatorcontrolled measurement of lung density can vary by up to 15 HU for regional measurements, and poor patient co-operation has a large effect. Semiautomatic methods can improve reproducibility although density estimates are greater because of partial volume effects.20 Regional changes in radiographic density of the lung in association with abnormalities of regional ventilation have been described since the 1930s, and changes in direct x-ray density during the breathing cycle were related to local pathology in the 1960s.23 Although CT studies of changes in lung tissue during respiration are limited, because of the radiation burden, there are reports of changes in lung properties with changes in lung volume in normal subjects,45 in patients during anaesthesia14and in intensive care.5 REGIONAL DIFFERENCES

How can the regional differences in radiographic density be related to lung expansion, and changes in radiographic density related to regional ventilation? First, we must establish a logical way to express regional ventilation. When a whole excised lung is inflated, expansion is relatively uniform. In an excised lung, therefore, lung volume could be expressed in terms of a proportion of the volume of the lung at any value convenient position on its pressure–volume characteristic. The most logical volume to use is total lung capacity (TLC). The degree of expansion of the lung is expressed as a fraction of the volume it would occupy at TLC. This

British Journal of Anaesthesia approach is necessary because in vivo, the lung is often not uniformly expanded. This method takes into account regional variations in expansion, by expressing the volume of a lung region as a proportion of the volume it can potentially occupy.32 As lung tissue is almost uniformly distended at TLC, even within the thorax, this allows an expression of ventilation as if it were “per alveolus”. Although this approach allows expression of the relative expansion of the lung regions, it requires knowledge of the volume of that region at TLC. When regional expansion is measured by the now classical technique of the inhaled radioactive isotope 133Xe, using external scintillation counters, this knowledge can be obtained from the counts taken during a breath-hold at TLC, with a uniform lung gas tracer concentration. In this way, having measured regional lung volume, regional differences in ventilation can be subsequently expressed in terms of “ventilation per maximal lung volume for that region”.31 Clearly, if different denominators were to be used to express the expansion of different lung regions, we would obtain illogical results. For example, consider a poorly expanded lung region that is at 10% of its maximal volume and is then expanded to 20% of its maximal volume. If we were to express its expansion in terms of its starting volume, then it would appear to have been expanded by 100%, whereas it has expanded by only 10% of its potential total volume. Therefore, if we are to use radiographic density changes to assess regional expansion and ventilation, we should not consider changes in density in terms of starting density, but rather in terms of the density of the region of interest at full lung expansion. Denison, Morgan and Millar measured the density of the lung at full expansion.3 They compared CT measurements of the lung with measurements of a foam-like model surrounded in a body-like shell to simulate the effects of beam hardening. They used bread with a range of consistencies—from moist pumpernickel to dry baguette—to simulate lung densities in conditions from fibrosis to emphysema. They were able to estimate density of the model substances accurately with CT and also measure precisely a volume of water injected into the bread. Normal lungs in supine subjects had a tissue volume of approximately 0.12 ml per ml of total volume at total lung capacity. They noted that in supine subjects, their measurements of density increased at both the apex and base of the lung, and suggested that this was the result of partial volume effects which makes measurements in these lung regions unreliable. On the basis of their estimates of density we can predict the relationship between lung expansion, expressed as a fraction of total lung capacity, physical density in g ml91, and radiographic density in HU. At total lung capacity (TLC), lung density DTLC:lung tissue mass/TLC. At any smaller lung volume (VL), lung density DV :lung tissue mass/VL. If the lung tissue mass is not affected by the change in volume, then fractional lung expansion VL/TCL:DTCL/DV . (This may not always be a valid assumption as some of the tissue mass consists of blood volume which may alter with expansion.) From the previous definition of the Hounsfield unit, radiographic density (Hounsfield units): L

L

Computed tomography and pulmonary measurements

Figure 1 Relationship between regional volume and radiographic density, for a lung density of 0.12 g cm93 at total lung capacity (TLC).

(DV 1000)91000. This relationship for the lung is shown in figure 1. The practical consequences of this reciprocal relationship are illustrated by the example in figure 2. Lung tissue at FRC is not homogeneously distended. The less well expanded lung has a greater density, as the same mass of tissue occupies a smaller volume. If during an inspiration, ventilation distribution throughout the lung is uniform (in proportion to the volume at TLC) then the less well expanded regions will have a greater change in radiological density. What components of the lung contribute to radiological density? In normal lungs, these are alveolar gas, lung tissue and blood in the capillaries. Increasing distension of the lung during inspiration leads to a progressive reduction in radiological density of the entire lung.4 45 46 55 There is a vertical gradient in lung density which can be related, at least in part, to the gradient in regional expansion. This gradient L

667 becomes less as lung volume increases towards TLC.45 46 55 57 The distribution of density values in the entire lung, which is unimodal during maximal inspiration, becomes bimodal during expiration, an effect which is seen only in the dependent parts of the lung.46 55 In animal studies, CT density was relatively constant for any particular vertical height, but CT measurements of lung gas volume systematically underestimated gas volume measured by helium dilution.54 Some of this systematic error is caused by partial volume effects. In lung disease such as emphysema where the relative volume of gas is increased30 or fibrosis where the tissue volume is greater,13 changes in CT density can be related to clinical, physiological, and pathological features. In abnormal lungs, there may also be a considerable amount of extravascular water in the form of oedema, either interstitial or in the airspaces. For example, haemodialysis can reduce overall lung density, both in aerated regions with densities of 9350 HU, and also in less well aerated regions with densities of 950 HU.55 When experimental pulmonary oedema is caused by increasing left atrial pressure, lung density increases by approximately 70%. The increase is more pronounced in central and dependent lung regions.18 In patients with chronic heart disease, where pulmonary vascular hypertrophy is likely, lung density measurements with CT can be related to the degree of pulmonary hypertension.53 In addition, it is clear from early x-ray density studies that the fluctuations in pulmonary artery pressure and x-ray density are closely related, indicating that changes in the volume of pulmonary vascular structures within the lung tissue have an important effect on radiological density.25 A valuable study used positron tomography to provide separate estimates of the tissue and blood components in suspine subjects, breathing at normal tidal volumes. Measurements were made over 15-min

Figure 2 Illustration of changes in density with uniform regional ventilation. At functional residual capacity (FRC), lung tissue is less expanded in dependent regions (at 40% of the volume at total lung capacity (TCL)) than in the upper regions (70% of TLC volume). These regions still consist of the same mass of tissue that they had at TLC, but they now fill less volume: for example, the density of the lower region is now 100/40 greater than at TLC. If the lung then expands uniformly by 20% of TLC volume, the relative change in density is far greater in the dependent region.

668

Figure 3 Predicted (lines) and observed (symbols) values for lung density, at total lung capacity (TLC) and residual volume (RV). Data are re-plotted from Millar and Denison.33 Note the deviation of the observed densities, indicating that density is affected by an additional gradient, probably vascular tissue volume.

period. This method did not have very good spatial resolution (1.6 cm) and therefore partial volume effects from overlapping of different tissues were great. Nevertheless, by using both transmission tomography with a source of 68Ge, and emission tomography after labelling blood with 11C carbon monoxide, the separate contributions of lung tissue density and blood density were separated in five volunteers. An anteroposterior gradient of lung density was found from approximately 0.2 g cm93 in the uppermost to approximately 0.4 g cm93 in the lowermost region. Mean lung density was 0.29 g cm93. Most of this gravitational gradient in density was attributable to blood and only a small amount to extravascular structures, considered to mostly represent lung tissue.44 Such observations are consistent with direct measurements of capillary blood and tissue expansion in animal studies8 9 and are supported by CT studies in dogs,54 and observations by CT studies in humans. Such studies show a gradient in lung density from ventral to dorsal regions, which is reduced but not abolished on inspiration.33 45 The gradient in density is reduced as transpulmonary pressure increases, but is still prominent at a transpulmonary pressure of 15 cm H2O and is not affected by voluntary inspiration that increases transpulmonary pressure to 20 cm H2O.57 Such findings should be compared with measurements of the gradient of alveolar volume by radioactive gas tracer methods, which show that the alveoli become progressively more uniformly expanded as overall lung volume increases, and hence the gravitational gradient in alveolar size becomes progressively less with increasing transpulmonary pressure.31 The discrepancies between the vertical distribution of alveolar volume, uniform at TLC, and the vertical distribution of radiological density, which persists at TLC, show clearly that radiological density does not merely measure “aeration” but a combination of aeration, blood volume and interstitial fluid volume. As blood has a much greater density, small changes in blood volume can affect radiological density far more than expansion or contraction of aerated lung tissue. For example, a Valsalva manoeuvre can decrese lung radiological density, particularly in

British Journal of Anaesthesia more dense regions, although lung volume and alveolar expansion are not affected.46 If we wish to relate accurately changes in lung radiological density to changes in alveolar volume (i.e. to infer changes in regional expansion or regional ventilation) then this is possible only if we know that the blood volume of the region remains unaltered. This may be a reasonable assumption for one gravitational level in the lung. If we compare different levels, then it must be remembered that absolute densities at TLC are different. Millar and Denison33 conducted an elegant study of the vertical gradient in CT density of the lungs of normal subjects. They found a discrepancy between the theoretical density of the lung, calculated on the basis of a simple hydraulic gradient of pressure in lung tissue, and the actual radiological density of the lung (fig. 3). They suggested that the discrepancy was because blood vessels increased in size in more dependent parts of the lung, imposing a second gradient in the tissue mass that contributes to radiological density. This effect is not relevant when regional volume change is considered on the basis of volume/maximal volume, as the tissue mass plays a negligible role in this estimate. These considerations indicate that the vertical gradient in CT density is likely to be the result of two factors: distribution of blood volume and distribution of lung expansion. The relative importance of these factors is not clear. How much these factors interact is also unclear, and such interactions may also be affected by mechanical ventilation, changes in circulating lung volume, pulmonary haemodynamics or pulmonary vascular disease. Lung density changes with position,7 vertical gradient36 and positive end-expiratory pressure5 have been studied in normal subjects and patients with ARDS in a series of studies by Gattinoni and his team, which have been summarized recently.35 They computed the regional distribution of tidal volume, but expressed this as a fraction of the volume of the lung section measured, not as a fraction of lung volume at TLC.6 This would not have been possible in patients with severe oedema and other pathological changes in the lung, as the ratio of gas to total weight, when lung tissue is fully distended, is unknown. They found an increase in vertical density gradient in patients with ARDS. Such a density gradient has been reported in alveolar oedema, but not in interstitial oedema.58 They found that positive endexpiratory pressure allowed dependent lung regions to remain distended and thus the distribution of ventilation became more homogenous. A potential source of error that these workers did not consider is the possibility that when alveolar pressure was increased, blood volume in capillaries could have been reduced, and this could decrease CT density. Such a decrease in density would, on the basis of their analysis, be interpreted as an increase in gas volume, and hence ventilation. For example, they reported that by increasing values of PEEP over a threshold value, there was a decrease in density in some lung slices, particularly the dependent ones. They attributed this to the pressure necessary to open compressed lung units, but a similar change could occur if the pressure were reducing tissue volume. Puybasset and co-workers have also used CT measurements in patients with ARDS to assess the

Computed tomography and pulmonary measurements

Figure 4 Relationship between radiographic density and regional volume for lung regions in the upper region with a density of 0.11 g cm93 (closed symbols) and dependent regions with a density of 0.15 g cm93 (open symbols), indicating the possible range of volume during spontaneous ventilation from functional residual capacity (FRC) to FRC and tidal volume (FRC;VT) (data from Reber and colleagues41).

distribution of disease,39 anatomical changes38 and aeration.40 Hedenstierna and co-workers, in a series of studies over many years,2 10–12 14–17 21 22 27 28 41 42 47 48 50 51 have studied the development of apparent atelectasis (“contrasting densities”) in subjects during anaesthesia and after abdominal surgery. Small patches of increased density developed during anaesthesia in dependent lung regions, and would move if the position of the subject changed. These appearances could be prevented or reduced by selective inflation of the dependent lung if the subject was in the lateral position,22 or by phrenic nerve stimulation in the supine subject.16 The increased density was related to gas exchange abnormality, particularly shunt, measured by the inert gas infusion method. However, the exact relationship between these areas of increased density and overall gas exchange impairment was not clear. For example, gas exchange efficiency during anaesthesia decreases with age, associated with an increase in lung regions with a small V/Q ratio, whereas the amount of increased density found on CT in older subjects was no greater than in younger subjects, and there was no relationship between shunt and age. Nevertheless, areas with increased radiological density on CT corresponded to areas of low ventilation/perfusion detected by independent methods of measuring distribution of ventilation and blood flow. In some circumstances, such as ketamine anaesthesia, in which lung volume did not decrease after induction of anaesthesia, such CT changes were not detected.52 The areas of increased density appeared to be made worse by 100% oxygen: subjects given 100% oxygen for 3 min before induction of anaesthesia had both a greater shunt and more radiological opacity.42 Although these workers generally used a visual method to identify areas of atelectasis in a single transverse CT image, usually taken just above the diaphragm, they subsequently showed that this particular region may not be greatly representative of the lung as a whole. For example, in nine anaesthetized supine patients, the mean distribution of blood flow to the most caudal CT slice was less than 10%, in contrast with the 70% for two slices that

669 were placed closer to the carina.51 Retrospective analysis of their radiological data showed that in awake subjects there was a single population of densities of lung voxels, with a mean density of approximately 9790 HU. After induction of anaesthesia, a small separate population of voxel densities was commonly found with densities of 9100 to ;100, although in some subjects the density of this second population was less, with densities in the range 9300 to ;50.29 Previous workers have reported a population of voxels with a greater density, of about this range of values, that became evident in expiration in normal conscious subjects.46 55 The 95% confidence limits for estimates of areas of increased density were more than 1%, and mean area was 3.7%, so the precision of the previous method of visual identification is not very great. This poor precision was often not apparent in previous studies where analysis had been made using regression methods only. Furthermore, in none of these studies does the assessment appear to have been performed using observers who were unaware of the state of the patient. The hypothesis generated by these measurements is that the association between lung volume decreases and the appearance of areas of atelectasis is causal.2 However, two similar studies do not directly support this causal link. First, CT studies in dogs during pentobarbital anaesthesia showed few and less marked densities.28 The authors expressed some reservations about these findings and suggested that these increased densities could have been caused by vascular congestion. As blood is probably the most important component of the radiographic density of normal lungs, even small increases in capillary volume increase density considerably. Second, when lung volume was reduced in conscious human subjects, by physical restriction and compression of the chest wall, no regions of increased density were noted. Values for mean radiographic density were not reported in this study, although these would have been of considerable interest.49 The same group of investigators reported changes in radiographic density in the entire lung, before and after induction of anaesthesia.41 Despite the fact that lung volume is usually reduced by anaesthesia, mean lung radiographic density did not alter. They also examined regional changes in lung density and compared awake and anaesthetized patterns, but in their analysis they assumed that gas distribution was proportional to changes in attenuation. If the changes in density with respiration are to be used to infer regional changes in volume, the regional density at TLC has to be known. Different values are reported, and density changes with the range of the vertical height of the lung because blood content is greater in dependent regions. However, a careful study of lung mechanics and CT changes in supine subjects reported that near TLC, the upper and lower parts of the lung have densities of approximately 9890 HU and 9850 HU, respectively.57 Based on the relationship derived above, this gives the relationships between ventilation and radiological density, for upper and lower parts of the lung, shown in figure 4. Measurements with 133Xe showed that the regional volumes of the upper and lower parts of the lung in a

670

British Journal of Anaesthesia cumstances, so that studies are limited to those where one or two “snapshots” of lung density can be rewarding.

Acknowledgements I thank Professors G. Hedenstierna and W. MacNee for help in the preparation of this article.

References

Figure 5 Regional ventilation in the conscious subject (top) and during anaesthesia and mechanical ventilation (bottom) for lung levels from upper (ventral—level 1) to lower (dorsal—level 5). Closed symbols indicate values for expiration and open symbols inspiration. The horizontal distance between the symbols therefore represents regional ventilation. Data calculated from mean values given in Reber and colleagues.41

supine subject are approximately 35% and 55% of the volume at TLC31 and these correspond with the CT density measurements at FRC reported by Reber and colleagues.41 Figure 4 indicates that at these volumes there are obvious differences in slope of the CT density–regional volume plot, between upper and lower zones. Despite the CT density change being approximately twice as large in the dependent slices, it is clear that the reported changes in CT density represent similar changes in regional volume for the upper and lower parts of the lung. If we assume a uniform change in lung density from upper to dependent regions, from 0.11 to 0.15, we can infer regional ventilation using the mean radiographic density data provided by Reber and co-workers41 (fig. 5). The distribution of ventilation is almost uniform in conscious subjects but becomes much more distributed to the upper zones after induction of anaesthesia and mechanical ventilation. These inferences correspond with measurements of diaphragm images made with a multiple CT method24 and earlier tracer studies.43 However, when the volume change for tidal breath during mechanical ventilation is calculated, the value obtained is less than during spontaneous ventilation. This is unlikely, because generally tidal volume is greater during mechanical ventilation. There are several possible explanations. First, the behaviour of this slice may not be the same during spontaneous and mechanical ventilation modes. The ribcage moves more during mechanical ventilation, and regional lung expansion may be altered between the two types of ventilation. More likely, there may be differences in pulmonary capillary blood volume, particularly differences in the shifts of pulmonary capillary blood with respiration, which could explain this discrepancy. In summary, if appropriate consideration is given to the characteristics of absorption, measurements of radiographic density of the lung allow useful inferences to be made about the distribution of gas, blood and tissue volume. However, comparisons are difficult when more than one component changes simultaneously. The measurements have some inaccuracies, particularly partial volume effects. Finally, radiation exposure limits the use in experimental cir-

1. Breinman RS, Beck JW, Korobkin M, Glenny R, Akwan OE, Heaston DK, Moore AV, Ram PC. Volume determinations using computed tomography. American Journal of Radiology 1982; 138: 329–333. 2. Brismar B, Hedenstierna G, Lundquist H, Strandberg A, Svensson L, Tokics L. Pulmonary densities during anesthesia with muscular relaxation—A proposal of atelectasis. Anesthesiology 1985; 62: 422–428. 3. Denison DM, Morgan MDL, Millar AB. Estimation of regional gas and tissue volumes of the lung in supine man using computed tomography. Thorax 1986; 41: 620–628. 4. Fromson BH, Denison DM. Quantitative features in the computed tomography of healthy lungs. Thorax 1988; 43: 120–126. 5. Gattinoni L, D’Andrea L, Pelosi P, Vitale G, Pesenti A, Fumigalli R. Regional effects and mechanism of positive end-expiratory pressure in early adult respiratory-distress syndrome. Journal of the American Medical Association 1993; 269: 2122–2127. 6. Gattinoni L, Pelosi P, Crotti S, Valenza F. Effects of positive end-expiratory pressure on regional distribution of tidal volume and recruitment in adult respiratory distress syndrome. American Journal of Respiratory and Critical Care Medicine 1995; 151: 1807–1814. 7. Gattinoni L, Pelosi P, Vitale G, Pesenti A, D’Andrea L, Mascheroni D. Body position changes redistribute lung computed-tomographic density in patients with acute respiratory failure. Anesthesiology 1991; 74: 15–23. 8. Glazier JB, Hughes JMB, Maloney JE, West JB. Vertical gradient of alveolar size in lungs of dogs frozen intact. Journal of Applied Physiology 1967; 23: 694–705. 9. Glazier JB, Hughes JMB, Maloney JE, West JB. Measurement of capillary dimensions and blood volume in rapidly frozen lungs. Journal of Applied Physiology 1969; 26: 65–76. 10. Gunnarsson L, Strandberg A, Brismar B, Tokics L, Lundquist H, Hedenstierna G. Atelectasis and gas-exchange impairment during enflurane–nitrous oxide anaesthesia. Acta Anaesthesiologica Scandinavica 1989; 33: 629–637. 11. Gunnarsson L, Tokics L, Gustavsson H, Hedenstierna G. Influence of age on atelectasis formation and gas-exchange impairment during general anaesthesia. British Journal of Anaesthesia 1991; 66: 423–432. 12. Gunnarsson L, Tokics L, Lundquist H, Brismar B, Strandberg A, Berg B, Hedenstierna G. Chronic obstructive pulmonary disease and anesthesia—formation of atelectasis and gasexchange impairment. European Respiratory Journal 1991; 4: 1106–1116. 13. Hartley PG, Galvin JR, Hunninghake GW, Merchant JA, Yagla SJ, Speakman SB, Schwartz DA. High-resolution CT-derived measures of lung density are valid indexes of interstitial lung disease. Journal of Applied Physiology 1994; 76: 271–277. 14. Hedenstierna G. Gas exchange during anaesthesia. British Journal of Anaesthesia 1990; 64: 507–514. 15. Hedenstierna G, Lundquist H, Lundh B, Tokics L, Strandberg A, Brismar B, Frostell C. Pulmonary densities during anaesthesia—an experimental study on lung morphology and gas-exchange. European Respiratory Journal 1989; 2: 528–535. 16. Hedenstierna G, Tokics L, Lundquist H, Adresson T, Strandberg A, Brismar B. Phrenic nerve stimulation during halothane anesthesia. Anesthesiology 1994; 80: 751–760. 17. Hedenstierna G, Tokics L, Strandberg A, Lundquist H, Brismar B. Correlation of gas exchange impairment to development of atelectasis during anaesthesia and muscle paralysis. Acta Anaesthesiologica Scandinavica 1986; 30: 183–191. 18. Hedlung LW, Vock P, Effmann EL, Lischko MM, Putman CE. Hydrostatic pulmonary edema. An analysis of lung density changes by computed tomography. Investigative Radiology 1984; 19: 254–262.

Computed tomography and pulmonary measurements 19. Hounsfield GN. Computerized transverse axial scanning (tomography): Part 1. Description of system. British Journal of Radiology 1973; 46: 1016–1022. 20. Kalender WA, Fichte H, Bautz W, Skalej M. Semiautomatic evaluation procedures for quantitative CT of the lung. Journal of Computer Assisted Tomography 1991; 15: 248–255. 21. Klingstedt C, Hedenstierna G, Baehrendtz S, Lundqvist H, Strandberg A, Tokics L, Brismar B. Ventilation–perfusion relationships and atelectasis formation in the supine and lateral positions during conventional mechanical and differential ventilation. Acta Anaesthesiologica Scandinavica 1990; 34: 421–429. 22. Klingstedt C, Hedenstierna G, Lundquist H, Strandberg A, Tokics L, Brismar B. The influence of body position and differential ventilation on lung dimensions and atelectasis formation in anaesthetized man. Acta Anaesthesiologica Scandinavica 1990; 34: 315–322. 23. Kourilsky R, Marchal M, Marchal MT. A new method of functional X-ray exploration of the lungs: photoelectric statidensigraphy. Diseases of the Chest 1962; 42: 345–358. 24. Krayer S, Rehder K, Vettermann J, Didier EP, Ritman EL. Position and motion of the human diaphragm during anesthesia–paralysis. Anesthesiology 1989; 70: 891–898. 25. Laws JW, Steiner RE. X-ray densitometry in the study of pulmonary ventilation and the pulmonary circulation. British Journal of Radiology 1965; 38: 512–519. 26. Levi C, Gray JE, McCullough EC, Hattery RR. The unreliability of CT numbers as absolute values. American Journal of Radiology 1982; 139: 443–447. 27. Lindberg P, Gunnarsson L, Tokics L, Secher E, Lundquist H, Brismar B, Hedenstierna G. Atelectasis and lung function in the postoperative period. Acta Anaesthesiologica Scandinavica 1992; 36: 546–553. 28. Lundquist H, Hedenstierna G, Ringertz H. Barbiturate anaesthesia does not cause pulmonary densities in dogs: a study using computerized axial tomography. Acta Anaesthesiologica Scandinavica 1988; 32: 162–165. 29. Lundquist H, Hedenstierna G, Strandberg A, Tokics L, Brismar B. CT-assessment of dependent lung densities in man during general anesthesia. Acta Radiologica 1995; 36: 626–632. 30. Macnee W, Gould G, Lamb D. Quantifying emphysema by CT scanning. Clinicopathologic correlates. Annals of the New York Academy of Sciences 1991; 624: 179–194. 31. Milic-Emili J, Crystal RG, ed. The Lung: Scientific Foundations, vol. 5. New York: Raven Press, 1991; 1043–1051. 32. Milic-Emili J, Henderson JAM, Dolovich MB, Trop D, Kaneko K. Regional distribution of inspired gas in the lung. Journal of Applied Physiology 1966; 21: 749–759. 33. Millar AB, Denison DM. Vertical gradients of lung density in healthy supine men. Thorax 1989; 44: 485–490. 34. Mull RT. Mass estimates by computed tomography: physical density from CT numbers. American Journal of Roentgenology 1984; 143: 1101–1104. 35. Pelosi P, Crotti S, Brazzi L, Gattinoni L. Computed tomography in adult respiratory distress syndrome: what has it taught us? European Respiratory Journal 1996; 9: 1055–1062. 36. Pelosi P, D’Andrea L, Vitale G, Pesenti A, Gattinoni L. Vertical gradient of regional lung inflation in adult respiratory distress syndrome. American Journal of Respiratory and Critical Care Medicine 1994; 149: 8–13. 37. Phelps ME, Gado MH, Hoffman EJ. Correlation of effective atomic number and electron density with attenuation coefficients measured with polychromatic X rays. Radiology 1975; 117: 585–588. 38. Puybasset L, Chaw N, Stewart T, Slutsky A, Lawkoune JD, Lu Q, Coriat P, Rouby JJ. Decrease in lung volume of consolidated pulmonary lower lobes prevents PEEP-induced alveolar recruitment in ARDS. British Journal of Anaesthesia 1996; 76: A144. 39. Puybasset L, Cluzel P, Gallart L, Rao GSU, Lawkoune JD, Lu Q, Slutsky A, Coriat P, Rouby JJ. CT scan assessment of the segmental distribution of ARDS-induced lung consolidation. British Journal of Anaesthesia 1996; 76: A142.

671 40. Puybasset L, Cluzel P, Rao GSU, Gallart L, Lawkoune JD, Slutsky A, Coriat P, Rouby JJ. CT scan assessment of PEEPinduced alveolar recruitment and overdistention. British Journal of Anaesthesia 1996; 76: A143. 41. Reber A, Engberg G, Sporre B, Kviele L, Rothen H-U, Wegenius G, Nylund U, Hedenstierna G. Volumetric analysis of aeration in the lungs during general anaesthesia. British Journal of Anaesthesia 1996; 76: 760–766. 42. Reber A, Engberg G, Wegenius G, Hedenstierna G. Lung aeration—the effect of pre-oxygenation and hyperoxygenation during total intravenous anaesthesia. Anaesthesia 1996; 51: 733–737. 43. Rehder K, Sessler AD, Rodarte JR. Regional intrapulmonary gas distribution in awake and anesthetized–paralyzed man. Journal of Applied Physiology 1977; 42: 391–402. 44. Rhodes CG, Wollmer P, Fazio F, Jones T. Quantitative measurement of regional extravascular lung density using positron emission and transmission tomography. Journal of Computer Assisted Tomography 1981; 5: 783–791. 45. Robinson PJ, Kreel L. Pulmonary tissue attenuation with computed tomography: comparison of inspiration and expiration scans. Journal of Computer Assisted Tomography 1979; 3: 740–748. 46. Rosenblum LJ, Mauceri RA, Wellenstein DE, Thomas FD, Bassano DA, Raasch BN, Chamberlain CC, Heltzman ER. Density patterns in the normal lung as determined by computed tomography. Radiology 1980; 137: 409–416. 47. Rothen HU, Sporre B, Engberg G, Wegenius G, Hedenstierna G. Re-expansion of atelectasis during general anaesthesia: a computed tomography study. British Journal of Anaesthesia 1993; 71: 788–795. 48. Strandberg A, Tokics L, Brismar B, Lundquist H, Hedenstierna G. Atelectasis during anaesthesia and in the postoperative period. Acta Anaesthesiologica Scandinavica 1986; 30: 154–158. 49. Tokics L, Hedenstierna G, Brismar B, Strandberg A, Lundquist H. Thoracoabdominal restriction in supine men: CT and lung function measurements. Journal of Applied Physiology 1988; 64: 599–604. 50. Tokics L, Hedenstierna G, Strandberg A, Brismar B, Lundquist H. Lung collapse and gas exchange during general anesthesia: effects of spontaneous breathing, muscle paralysis, and positive end-expiratory pressure. Anesthesiology 1987; 66: 157–167. 51. Tokics L, Hedenstierna G, Svensson L, Brismar B, Cederlund T, Lundquist H, Strandberg A. V/Q distribution and correlation to atelectasis in anesthetized paralyzed humans. Journal of Applied Physiology 1996; 81: 1822–1833. 52. Tokics L, Strandberg A, Brismar B, Lundquist H, Hedenstierna G. Computerized tomography of the chest and gas exchange measurements during ketamine anaesthesia. Acta Anaesthesiologica Scandinavica 1987; 32: 684–692. 53. Vock P, Salzmann CH. Comparison of computed tomographic lung density with haemodynamic data of the pulmonary circulation. Clinical Radiology 1986; 37: 459–464. 54. Wandtke JC, Hyde RW, Fahey JP. Measurement of lung gas volume and regional density by computed tomography in dogs. Investigative Radiology 1986; 21: 108–117. 55. Wegener OH, Koeppe P, Oeser H. Measurement of lung density by computed tomography. Journal of Computer Assisted Tomography 1978; 2: 263–273. 56. Williams G, Bydder GM, Kreel L. The validity and use of computed tomography attenuation values. British Medical Bulletin 1980; 36: 279–287. 57. Wollmer P, Albrechtsson U, Brauer K, Eriksson L, Jonson B, Tylen U. Measurement of pulmonary density by means of x-ray computerized tomography. Chest 1986; 90: 387–391. 58. Wollmer P, Rhodes CG, Deanfield J, Rozkovec A, Allan RM, Maseri A, Fazio F. Regional extravascular density of the lung in patients with acute pulmonary oedema. Journal of Applied Physiology 1987; 63: 1890–1895.