Background: In acute lung injury, thoracic CT is used to gain information about lung aeration and consolidation. This can be done either during breath-holding by spiral CT scanning of the entire lung or dynamically by scanning lung slices without interrupting ventilation. We hypothesized that attenuation distribution is dependent on static or dynamic scanning techniques. We also studied whether a variation in the CT cut level, corresponding to the diaphragm movement over a breath, had any effect on the attenuation distribution.
Methods: Twenty-two pigs with oleic acid-induced lung injury were randomly assigned to receive pressure-controlled mechanical ventilation with or without spontaneous breathing. Transversal dynamic CT scans of the chest were performed in apical and juxtadiaphragmatic regions, and end-expiratory and end-inspiratory slices were selected. In addition, after clamping the tube at end-expiration and end-inspiration, respectively, spiral CTs were performed. Guided by morphologic structures, spiral CT slices matching the dynamic scan slice and three additional neighbored slices above the diaphragm were selected. Distributions of CT attenuation were calculated and summarized in ranges for comparison.
Results: No significant difference in attenuation distributions between the two scanning methods or an interaction with the factors ventilation mode, ventilation phase, and attenuation range were found. In addition, attenuation distributions of four neighbored juxtadiaphragmatic slices, 8 mm thick, from the spiral CT did not differ statistically.
Conclusion: In an animal model of oleic acid-induced lung injury, analyses of transverse thoracic slices based on dynamic or static CT scanning showed comparable distributions of attenuation. Variations on the CT cut level of 24 mm had no significant effect on the distribution of Hounsfield unit numbers. CT attenuation distributions of transversal juxtadiaphragmatic slices were not dependent on exact position.
Key words: acute lung injury; CT; lung aeration; lung consolidation
Abbreviations: ALI = acute lung injury; APRV = airway pressure-release ventilation; HU = Hounsfield unit; n.s. = not significant; PCV = pressure-controlled ventilation; PEEP = positive end-expiratory pressure; ROI = region of interest; RR = respiratory rate; SB = spontaneous breathing
CT scanning is used in acute lung injury (ALI) to gain information about lung aeration and consolidation as well as effects of mechanical ventilation. (1,2) With the evolution of CT scanners, two basic protocols have been used to study lung CT attenuation as measured in Hounsfield units (HU). A traditional approach uses one or several single slices to dynamically scan a lung region during the ventilatory cycle and limit radiation exposure. (3,4) A newer approach allows whole lung reconstruction from a fast spiral CT scan while clamping the tube and thereby interrupting ventilation.5 However, CT scan results are different between end-expiration and end-inspiration, (2,6) and redistribution of inhomogeneously distributed air may influence results during tube clamping (lasting up to > 10 s). This might especially affect high-resolution spiral CT comparisons with respect to regional density distributions of attenuation between different ventilation settings. (7-9) Differences between ventilation modes might even be enhanced in presence of spontaneous breathing (SB), as diaphragmatic activity during the clamping phase may favor redistribution of gas within the lung.
Fast CT scanners, however, allow dynamic scanning with multiple revolutions of the radiograph tube. Images can be reconstructed for one slice with temporal resolutions of typically 0.1 to 0.25 s with and without interrupting ventilation. (10-15)
Controlled ventilation is as widely used as ventilation modes with SB, and it can be assumed that comparisons between the aforementioned CT scanning techniques might be especially difficult during a ventilation mode with SB. It has also been claimed that scanning of a single transverse slice may not show a representative result for a certain lung region such as the juxtadiaphragmatic region. Differences between transverse regions are also discussed when a fixed position is scanned at end-inspiration and end-expiration because not necessarily the same lung tissue was scanned when CT attenuation distributions at end-inspiration and end-expiration are used for comparison. (16-19)
These methodologic issues were studied in a model of porcine experimental lung injury. We hypothesized the following: (1) dynamic scanning in one transverse slice during time-cycled pressure-controlled ventilation (PCV) with or without SB reveals different lung attenuation distributions when compared with static spiral CT scanning in the same slice with tube clamping and interruption of ventilation, and (2) that CT attenuation distributions will differ between four neighboring slices located above the diaphragm.
MATERIALS AND METHODS
Experimental Setting and Protocol
The study was performed in the research laboratory of the Department of Clinical Physiology at the University Hospital of Uppsala, Sweden, and was approved by the local animal ethics committee. Dynamic CT scans were performed in 22 pigs with oleic acid-induced lung injury. These pigs were also included in a porcine study of end-expiratory atelectasis and lung volume. (20) Anesthesia was induced IM and maintained by infusion of 30 mg/kg/h of ketamin, 0.1 mg/kg/h of midazolam, and 1 to 2 [micro]g/kg/min of remifentanil. After tracheotomy in supine position, the pigs received time-cycled PCV in the airway pressure-release ventilation (APRV) mode of a standard ventilator (Evita 4; Drager; Lubeck, Germany). Lung injury was induced with repeated central venous oleic acid injections (0.1 mL/kg). Two hours after induction of ALI, animals were randomized using sealed envelopes to either continue with PCV or to allow SB with APRV. PCV was applied with a respiratory rate (RR) of 20 breaths/min, an inspiratory/expiratory ratio of 1:1, a fraction of inspired oxygen of 0.5, a positive end-expiratory pressure (PEEP) of 5 cm [H.sub.2]0, and an inspiratory pressure that resulted in a tidal volume of approximately 10 mL/kg. Adjustments of RR up to 30 breaths/min and increments of inspiratory pressure were allowed to avoid hypercapnia above a PaC[O.sub.2] of 60 mm Hg and to compensate for decreased compliance. To suppress SB during hypercapnia, remifentanil infusion was increased to 2 [micro]g/kg/min and, if necessary, 2.5 mg/h of pancuronium bromide was infused for muscle relaxation. To allow reinstitution of SB in the APRV group, RR was decreased to 15 breaths/min, resulting in (preset) inspiratory and expiratory times of 2 s, respectively, and remifentanil infusion was lowered. PEEP, inspiratory/expiratory ratio, and fraction of inspired oxygen were kept constant during the entire study. These ventilator settings resulted in comparable mean and end-inspiratory airway pressures, tidal and minute ventilation, and PaC[O.sub.2] whether SB was present or not, whereas RR was higher with SB. (20) Four hours after randomization, pigs were transferred to the radiology department without interrupting ventilation or changing ventilatory mode, and transverse dynamic chest CT scans were performed.
CT Scanning and Analysis
During ventilation, a frontal topogram of the chest was obtained with a Somatom Plus 4 CT Scanner (Siemens; Erlangen, Germany). A transversal scan (140 kilovolts, 111 mA) 1 to 2 cm above the diaphragm and one apical scan located at half the way of the intrathoracic trachea were acquired in the dynamic scanning mode (0.75 s for one complete revolution), with a collimation of 8 mm for an acquisition time of 4.5 s ensuring inclusion of at least one ventilatory cycle. The CT scanner provided image reconstructions with a slice thickness of 8 mm and effective time resolution of 0.1 s, resulting in 45 single scans for each slice with a matrix of 512 x 512 voxels.
Thereafter, guided by observation of airway pressure and flow curves, the tube was damped strictly at end-expiration followed by a spiral scan (140 kilovolts, 111 mA) with a collimation of 8 mm and a pitch of 1.5; transverse images were reconstructed with a slice thickness of 8 mm including all of both lungs. The craniocaudal scanning direction was randomized. This procedure was repeated with clamping the tube at end-inspiration.
CT images were transferred to a personal computer and analyzed with a computer program (Osiris; University of Geneva; Geneva, Switzerland). The investigator of CT images (T.M.) was specially trained and blinded to ventilatory mode. The following analysis was performed:
1. Guided by morphologic structures, the slice matching the dynamic transverse scans in the apical region and above the diaphragm were selected from the spiral CT for analysis of region of interest (ROI) [see step 3]. Above the diaphragm, three additional slices in cranial direction each 8 mm apart away from the previous scan were selected.
2. In the dynamic scans, the image with the highest mean HU and the lowest mean HU in the following sequence of images with decreasing HU numbers, representing the end-expiratory and end-inspiratory images, were selected by another person (J.Z.) from a roughly drawn preliminary ROI within the lungs in order not to unblind the ventilatory mode and passed to the investigator (T.M.) for analysis.
3. In all slices from static spiral and dynamic CT scanning, the entire left and right lungs were chosen as a ROI by drawing the external boundaries of the lungs at the inside of the ribs and the internal boundaries along the mediastinal organs. The number of pixels corresponding to each attenuation value in the ROI of each slice were counted and stored by the computer program. For repeated marking of the external boundaries of the lungs using similar technology, the intraobserver variability for the determination of ROIs was 1.7% of the mean ROI area. (21)
4. Continuous CT attenuation distributions for the selected images of dynamic scans at end-expiration and end-inspiration and matching slices from spiral CTs performed with clamped tube at end-expiration and end-inspiration were constructed. Attenuation distributions were also calculated for additional slices above the diaphragm that were located 8 mm apart from each other. In addition, four groups of attenuation ranges with decreasing air content were defined as described previously (22-25): range 1 included attenuation numbers between -1,000 HU and -900 HU, previously defined as hyperinflation; range 9, between -900 HU and -500 HU, defined as normal aeration; range 3 between -500 HU and -100 HU, previously defined as poor aeration; and range 4 between -100 HU and 100 HU, representing atelectasis or lung parenchyma with an air content of [less than or equal to] 10%. Attenuation values outside the range of -1,000 HU and + 100 HU, which constituted < 1% of all counts, were excluded.
Primary outcome measures were summed frequencies in the above defined HU ranges (hyperinflation, normal aeration, poor aeration, and atelectasis or lung parenchyma with an air content of [less than or equal to] 10%). Descriptive statistics for comparison of the scanning methods were performed according to Bland and Altman. (26)
Sample size estimation was based on findings of previous studies. (20) To detect differences in a HU range between ventilatory settings with the given two-sided parallel design at a significance level of 5% ([alpha] = 0.05) with a probability of 80% ([beta] = 0.20) based on an estimated difference of 0.76 of the mean SD of the parameter at least 20 animals had to be studied. Results are expressed as mean [+ or -] SD.
All statistical analyses were performed using a statistical software package (Statistica for Windows 6.0; StatSoft; Tulsa, OK). Normal distribution of data were confirmed by a Shapiro-Wilks W test, and data were analyzed using multifactor analysis of variance with factors CT scanning method, spiral CT scanning direction, ventilation mode, ventilation phase, and HU range (as repeated-measures factor). For separate analyses of slices above the diaphragm obtained from the spiral CT, an additional factor for the distance from the diaphragm was included. Differences were considered to be statistically significant at p < 0.05.
The continuous distributions of single-voxel attenuation for PCV are shown in Figure 1 for apex and diaphragm as well as for end-inspiration and end-expiration. No differences between the two scanning methods are apparent for the complete range of HUs. With SB present in the APRV mode, the continuous distribution of single-voxel attenuation in Figure 2 also exhibits no differences between the two scanning methods.
[FIGURES 1-2 OMITTED]
Statistical analysis showed no significant differences in any of the four analyzed ranges of HU (Fig 1, 2). The main factor, CT scanning method, was not significant (n.s.) in the diaphragmatic and in the apical slice, and no interaction with factors spiral CT scanning direction, ventilation phase, or HU range with the factor CT scanning method was detected. There was also no interaction for the factor ventilation mode, showing that the spontaneous breaths present in the APRV mode have no influence on the comparison of the CT scanning methods.
For comparison of single values for every class of HU of the continuous distribution plot (every data point), the difference of normalized frequency between the dynamic scanning method and the spiral CT was calculated in every pig and plotted against its mean value according to Bland and Altman (Fig 3). The agreement between the scanning methods is good, few outliers are present, but there is no apparent pattern in any of the plots. The mean of differences as measure for bias was lower than 1 x [10.sup.-7] for every plot; the 2-SD interval as measure for accuracy was 0.013 and 0.012 for the slices in the apex and diaphragm for APRV with SB and 0.014 and 0.013 for PCV, respectively (values for expiration and inspiration calculated together).
[FIGURE 3 OMITTED]
An additional three slices from the spiral CT further apart from the diaphragm did not differ statistically from the slice directly above the diaphragm. No difference was found for the factor distance from the diaphragm for any of the additional slices (Table 1).
Ventilatory variables during APRV with SB and PCV were similar: RR was 39 [+ or -] 9 breaths/rain and 34 [+ or -] 5 breaths/min; tidal volume was 234 [+ or -] 56 mL and 231 [+ or -] 48 mL; minute volume was 8.7 [+ or -] 1.6 L and 8.0 [+ or -] 1.6 L; PaC[O.sub.2] was 57 [+ or -] 16 mm Hg and 56 [+ or -] 15 mm Hg; end-inspiratory pressure was 24.8 [+ or -] 4.0 cm [H.sub.2]O and 24.0 [+ or -] 3.8 cm [H.sub.2]O; and mean airway pressure was 14.6 [+ or -] 2.5 cm [H.sub.2]O and 13.8 [+ or -] 2.3 cm [H.sub.2]O for APRV with SB and PCV, respectively. No statistical differences between the ventilatory variables were observed.
In an animal model of oleic acid-induced ALI, we found no differences between dynamic or static CT scanning methods studied. No effect of additional factors such as presence of SB, ventilation phase (inspiration or expiration), attenuation number range, or exact transversal scanning position on differences between distributions of radiologic densities for the two CT scanning methods was observed. These results were not expected because of the major differences in the applied CT scanning methods.
Dynamic CT scanning at a fixed scanning level does not interrupt the ongoing ventilation and should therefore result in scanning different lung tissue during the ventilation cycle. From the sequence of the reconstructed images, the end-inspiratory and end-expiratory images were selected, which fall into a phase of the ventilation cycle with small volume changes and therefore slow movement of the lung tissue. Comparability between the selected images from the dynamic scans and the spiral CT slices was secured by choosing the appropriate slice from the spiral CT from matching anatomic structures like large vessels and dimension of lung parts in both images. The resolution of 8 mm for the spiral CT in the cranial direction generates enough overlap of the images from dynamic and spiral CT in the image plane to ensure comparability.
The reconstruction algorithm for the dynamic CT allows an image frequency of 0.1 s, which is significantly shorter than a full scanner rotation and is explained by an overlap of data from one image to the next. This limitation is of minor importance for this study, because the time course of the images generated was only needed to select the end-inspiratory and end-expiratory images for the comparison with the static images from the spiral CT. An advantage of the dynamic scanning method with selection of end-inspiratory and end-expiratory images is that external triggering of the CT acquisition in order to start at a specific instant in the ventilatory cycle is not needed.
The acquisition of the spiral CT required closing of the airway by tube clamping for several seconds with the exact time depending on the dimension of the lungs at end-inspiration and end-expiration. This procedure is necessary to generate static conditions during the acquisition time for the spiral CT and is common practice also in patients if the whole lung is of interest for evaluation. (16,27) In inhomogeneously aerated lungs with ALI, we expected intrapulmonary gas redistribution during clamping, which was not observed in our study. Since the CT scanning was only performed at the end of the study period, no conclusions can be drawn about possibly different results during the time course of the study, but in our model of moderate lung injury gas redistribution should not have been inhibited completely and large loss of lung volume was only present in the PCV group, (20) while no differences between the ventilation modes were observed. There were no significant differences between the studied HU ranges and the continuous distributions of lung attenuation were very well matched throughout the whole range of HU numbers. The same was true for end-inspiratory and end-expiratory scans. Additionally, Bland and Altman plots of differences between the scanning methods against their means for classes of 20 HU for every pig separately demonstrate good agreement between the scanning methods and show that the scatter in the data should not have precluded finding differences between the CT scanning methods.
Although SB efforts during the clamping phase could have influenced gas redistribution within the lung, no such interaction was observed. Ventilatory variables for PCV and APRV with SB were similar so that a difference for the factor ventilation mode would have been expected to result mainly from SB. The factor scanning direction for the spiral CT (starting at the apex or base) did not modify the comparison of the scanning methods for the apex and diaphragm analyzed separately, showing that the time elapsed from the beginning of the clamping until the lung slice is finally scanned during the spiral CT is of minor importance, even though the clamping phase can last several seconds. It is important to remember that differences between the distribution of HU units between APRV with SB and PCV in the apex and the four slices in the diaphragmatic region emerge from the fact that two different groups of animals were treated with these two ventilation modes, although ventilatory variables were not significantly different.
Three additional slices above the diaphragm selected from the static spiral CT, each 8 mm apart, showed no statistically significant differences for any of the compared ranges of HU. This finding shows that the above results for the comparison of scanning methods is not dependent on the exact position of the single slice selected for the dynamic CT scan, the result is not modified by the additional factors ventilation mode and ventilation phase. This finding and the close matching of spiral CT and the dynamic CT in one slice also suggests that no differences for the dynamic CT scan should be expected in the four slices above the diaphragm. The observations about additional slices of lung tissue are only true for a limited small range of 2.4 cm above the diaphragm in this case, as the distribution of aerated and collapsed tissue can vary markedly through the whole lung. (16,28)
Previous studies with fast dynamic CT scanning showed that reconstruction of images every 0.1 to 0.25 s can be used to study attenuation distributions in a single slice to assess changes during the ventilation cycle (12,13) and correlation with physiologic variables in experimental lung injury. (29) Recent work (15) showed that in different HU ranges results from a single infracarinal slice scanned with dynamic CT did not correlate well with lung area from a whole spiral CT, especially in the range of -300 to 200 HU. This study compared results from a spiral CT for the whole lung with a dynamic CT in a single slice and does therefore not allow us to conclude whether the differences emerged from use of different scanning methods or from studying different parts of the lung. The fact that a single slice may not be representative for the whole lung was previously addressed with PEEP-induced changes in a study with spiral CT scanning. (16)
In summary, our study shows that in an animal model with ALI dynamic CT scanning for one transverse slice and the matching slice of a static spiral CT scan with clamped tube showed the same distribution of CT attenuation. This result was not influenced by the ventilation phase and the presence of SB. Three additional spiral CT slices above the diaphragm showed no differences for the distribution of voxels in the different HU ranges. Our data suggest that in a model of experimental lung injury, comparisons of CT attenuation distributions, eg, to analyze regional distribution of air and nonaerated tissue between different ventilatory modes or aeration changes during tidal ventilation, can be performed with dynamic as well as static CT scanning techniques in a fixed transversal slice without significant influence on the results.
ACKNOWLEDGMENT: We thank Ms. Eva-Maria Hedin, Ms. Anne Abrahamson, Ms. Agneta Roneus, all technicians at Department of Clinical Physiology, and the radiograph laboratory team (Ms. Marianne Almgren, Ms. Ann Erikson, Ms. Ewa Larsson, all technicians at Department of Radiology), University of Uppsala, Sweden; and Christian Rylander, Department of Anaesthesiology and Intensive Care, Sahlgrenska University Hospital, Goteborg, Sweden, for technical help
(1) Gattinoni L, Pesenti A, Torresin A, et al. Adult respiratory distress syndrome profiles by computed tomography. J Thorac Imaging 1986; 1:25-30
(2) Gattinoni L, Caironi P, Pelosi P, et al. What has computed tomography taught us about the acute respiratory distress syndrome? Am J Respir Crit Care Med 2001; 164:1701-1711
(3) Gattinoni L, Pelosi P, Vitale G, et al. Body position changes redistribute lung computed-tomographic density in patients with acute respiratory failure. Anesthesiology 1991; 74:15-23
(4) Gattinoni L, Mascheroni D, Torresin A, et al. Morphological response to positive end expiratory pressure in acute respiratory failure: computerized tomography study. Intensive Care Med 1986; 12:137-142
(5) Vieira SR, Puybasset L, Richecoeur J, et al. A lung computed tomographic assessment of positive end-expiratory pressure-induced lung overdistension. Am J Respir Crit Care Med 1998; 158:1571-1577
(6) Gattinoni L, Pelosi P, Crotti S, et al. Effects of positive end-expiratory pressure on regional distribution of tidal volume and recruitment in adult respiratory distress syndrome. Am J Respir Crit Care Med 1995; 151:1807-1814
(7) Puybasset L, Cluzel P, Gusman P, et al. Regional distribution of gas and tissue in acute respiratory distress syndrome: I. Consequences for lung morphology CT scan; ARDS Study Group. Intensive Care Med 2000; 26:857-869
(8) Puybasset L, Gusman P, Muller JC, et al. Regional distribution of gas and tissue in acute respiratory distress syndrome: III. Consequences for the effects of positive end-expiratory pressure CT scan; ARDS Study Group. Intensive Care Med 2000; 26:1215-1227
(9) Malbouisson LM, Muller JC, Constantin JM, et al. Computed tomography assessment of positive end-expiratory pressure-induced alveolar recruitment in patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 163:1444-1450
(10) Neumann P, Berglund JE, Mondejar EF, et al. Effect of different pressure levels on the dynamics of lung collapse and recruitment in oleic-acid-induced lung injury. Am J Respir Crit Care Med 1998; 158:1636-1643
(11) Bink A, Markstaller K, Birkenkamp K, et al. Multi-rotation CT and acute respiratory distress syndrome: animal experiment studies. Radiologe 2001; 41:195-200
(12) Markstaller K, Kauczor HU, Eberle B, et al. Multi-rotation CT during continuous ventilation: comparison of different density areas in healthy lungs and in the ARDS lavage model [in German]. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 1999; 170:575-580
(13) Markstaller K, Eberle B, Kauczor HU, et al. Temporal dynamics of lung aeration determined by dynamic CT in a porcine model of ARDS. Br J Anaesth 2001; 87:459-468
(14) Markstaller K, Karmrodt J, Doebrich M, et al. Dynamic computed tomography: a novel technique to study lung aeration and atelectasis formation during experimental CPR. Resuscitation 2002; 53:307-313
(15) Bletz C, Markstaller K, Karmrodt J, et al. Quantification of atelectases in artificial respiration: spiral-CT versus dynamic single-slice CT [in German]. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 2004; 176:409-416
(16) Lu Q, Malbouisson LM, Mourgeon E, et al. Assessment of PEEP-induced reopening of collapsed lung regions in acute lung injury: are one or three CT sections representative of the entire lung? Intensive Care Med 2001; 27:1504-1510
(17) Puybasset L, Cluzel P, Chao N, et al. A computed tomography scan assessment of regional lung volume in acute lung injury: the CT Scan ARDS Study Group. Am J Respir Crit Care Med 1998; 158:1644-1655
(18) Malbouisson LM, Muller JC, Constantin JM, et al. Computed tomography assessment of positive end-expiratory pressure-induced alveolar recruitment in patients with acute respiratory distress syndrome. Am J Respir Crit Care Med 2001; 163:1444-1450
(19) Drummond GB. Computed tomography and pulmonary measurements. Br J Anaesth 1998; 80:665-671
(20) Wrigge H, Zinserling J, Neumann P, et al. Spontaneous breathing improves lung aeration in oleic acid-induced lung injury. Anesthesiology 2003; 99:376-384
(21) Rylander C, Hogman M, Perchiazzi G, et al. Oleic acid lung injury: a morphometric analysis using computed tomography. Acta Anaesthesiol Scand 2004; 48:1123-1129
(22) Gattinoni L, Pelosi P, Vitale G, et al. Body position changes redistribute lung computed-tomographic density in patients with acute respiratory failure. Anesthesiology 1991; 74:15-23
(23) Gattinoni L, D'Andrea L, Pelosi P, et al. Regional effects and mechanism of positive end-expiratory pressure in early adult respiratory distress syndrome (published erratum appears in JAMA 1993; 270:1814) JAMA 1993; 269:2122-2127
(24) Vieira SR, Puybasset L, Richecoeur J, et al. A lung computed tomographic assessment of positive end-expiratory pressure-induced lung overdistension. Am J Respir Crit Care Med 1998; 158:1571-1577
(25) Lundquist H, Hedenstierna G, Strandberg A, et al. CT-assessment of dependent lung densities in man during general anaesthesia. Acta Radiol 1995; 36:626-632
(26) Bland JM. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307-310
(27) Puybasset L, Cluzel P, Chao N, et al. A computed tomograplay scan assessment of regional lung volume in acute lung injury: the CT Scan ARDS Study Group. Am J Respir Crit Care Med 1998; 158:1644-1655
(28) Gattinoni L, Presenti A, Torresin A, et al. Adult respiratory distress syndrome profiles by computed tomography. J Thorac Imaging 1986; 1:25-30
(29) Markstaller K, Kauczor HU, Weiler N, et al. Lung density distribution in dynamic CT correlates with oxygenation in ventilated pigs with lavage ARDS. Br J Anaesth 2003; 91: 699-708
Jorg Zinserling, MSc; Hermann Wrigge, MD; Peter Neumann, MD, PhD; Thomas Muders, MD; Anders Magnusson, MD, PhD; Goran Hedenstierna, MD, PhD; and Christian Putensen, MD, PhD
* From the Department of Anesthesiology and Intensive Care Medicine (Mr. Zinserling, and Drs. Wrigge, Muders, and Putensen), University of Bonn, Bonn, Germany; Department of Anesthesiology and Intensive Care Medicine (Dr. Neumann), University of Gottingen, Gottingen, Germany; and Departments of Radiology (Dr. Magnusson) and Clinical Physiology (Dr. Hedenstierua), University, of Uppsala, Uppsala, Sweden.
This study was supported by grants from Deutsche Forschungsgemeinschaft (DFG; Pu 219/1-1), Bonn, Germany; the Swedish Medical Research Council (no 5315), Stockholm, Sweden; and the Swedish Heart-Lung-Foundation, Stockholm, Sweden. Manuscript received January 25, 2005; revision accepted March 18, 2005.
Correspondence to: Jorg Zinserling, MSc, Department of Anesthesiology and Intensive Care Medicine, University of Bonn, D-53105, Bonn, Germany; e-mail email@example.com
COPYRIGHT 2005 American College of Chest Physicians
COPYRIGHT 2005 Gale Group