Atelectasis, or collapse of the alveoli and absence of air from the lungs,1 is important from a diagnostic imaging viewpoint because a large proportion of dogs that undergo CT are anesthetized and receive concurrent supplementary oxygen, both of which are major contributors to the development of pulmonary atelectasis1–6 and may mask or mimic lesions in the lungs.7 Furthermore, atelectasis may be exacerbated in patients that have been anesthetized and positioned in lateral recumbency for a period of time prior to CT examination.7
In human patients, atelectasis becomes evident on CT images within 5 minutes after induction of anesthesia.8 In anesthetized dogs, notable attenuation of lung tissue on CT images develops within 7 minutes after being positioned in lateral recumbency, and that attenuation resolves or returns to basal levels within 7 minutes after being repositioned in sternal recumbency.9 Supplementation with a high inspired oxygen fraction (80% to 100%), which is commonly administered concurrently with inhalation anesthetics,1 exacerbates the extent of atelectasis, compared with supplementation with lower inspired oxygen fractions (30% to 40%).2,4,5,10 Results of a study6 in which the extent of atelectasis was compared between sedated dogs positioned in lateral recumbency that were breathing room air and similar dogs that were breathing 100% oxygen indicate that oxygen supplementation exacerbates atelectasis. In another study,11 21 of 28 (75%) dogs anesthetized with sodium pentobarbital and mechanically ventilated with room air during CT examination did not develop evidence of atelectasis, and for the 7 dogs that did develop evidence of atelectasis, the extent of the pulmonary changes was very mild, compared with that reported for human patients. All those studies6,9,11 evaluated important factors in a clinical environment, specifically anesthesia or sedation, inspired gas composition or oxygen supplementation, and body position; however, results indicate that many of those factors are irrelevant from an imaging viewpoint when assessed in regard to modern helical CT scanners, modern anesthetic agents, and the detailed information available for affected lung lobes (eg, location within each lobe and time to development of atelectasis). Additionally, image quality will be compromised for patients with atelectasis prior to CT scanning, and clinicians need to decide whether to abort the scan or attempt to resolve the atelectasis before commencing the scan in those patients.
The aims of the study reported here were to characterize the extent and location of atelectasis in dogs anesthetized with a commonly used anesthetic protocol and positioned in lateral recumbency and to determine whether repositioning the dogs in sternal recumbency would resolve atelectasis. This study was clinically relevant because patients positioned in lateral recumbency are prone to the formation of atelectasis, which may compromise CT image quality and hinder diagnosis of pathological lesions. We hypothesized that the dependent lung lobes would be most susceptible to atelectasis, and on CT images, atelectasis would be characterized by a decrease in the cross-sectional area and an increase in the quantitative attenuation of the affected lobes.
Materials and Methods
Animals
The study was approved by the Animal Ethics Committee of the University of Pretoria. Additionally, radiation safety and As Low As Reasonably Achievable (ALARA) principles were adhered to throughout the study.
Six adult spayed female Beagles that belonged to the Onderstepoort Animal Teaching Unit were used for the study. The dogs had a median age of 8 years (range, 8 to 9 years) and median weight of 13 kg (range, 10.8 to 13.7 kg). All dogs were considered to be in good body condition (body condition score range, 5 to 6 as determined on a 9-point scale). Older (≥ 8 years) dogs were purposely selected for the study because the risk of radiation exposure is inherently greater for young dogs, compared with older dogs. Thus, the sample size was limited by the number of older dogs in the teaching unit.
All 6 dogs were determined to be healthy on the basis of results of a physical examination, CBC, abbreviated serum biochemical analysis (consisted of total serum protein, albumin, urea, creatinine, cholesterol, and glucose concentrations and alanine aminotransferase and alkaline phosphatase activities), and abdominal ultrasonographic examination. For each dog, lung lesions were ruled out by means of a survey CT scan (which was included in the study).
CT procedure
Each dog underwent 2 CT examinations; 1 was performed with the dog positioned in LLR, and 1 was performed with the dog positioned in RLR. There was a 2-week interval between examinations. The dogs were randomly allocated into 2 groups (3 dogs/group) by use of a coin-toss method. The dogs in group 1 were positioned in RLR for the first CT examination and LLR for the second examination, whereas the dogs in group 2 were positioned in LLR for the first CT examination and RLR for the second examination. Within each group, dogs were consecutively scanned on 1 day.
Dogs were anesthetized for each examination. Food but not water was withheld from dogs for 8 to 12 hours before anesthesia was induced. One dog was inadvertently fed within 8 hours of anesthesia induction, but that fact was not discovered until the baseline CT scan was initiated, and the decision was made to continue the examination. Each dog was anesthetized in accordance with the standard protocol for patients at the Onderstepoort Veterinary Academic Hospital, which included premedication with morphine sulfate (0.2 mg/kg, IM) 15 minutes before administration of diazepam (0.2 mg/kg, IV via a cephalic catheter). Dogs were positioned in sternal recumbency, and anesthesia was induced with propofol (4 to 6 mg/kg, IV to effect). After endotracheal intubation, anesthesia was maintained with oxygen (flow rate, 1 L/min) and isoflurane (2% to 2.5%). Dogs were not mechanically ventilated, and the end-tidal isoflurane concentration was maintained between 1.4% and 1.6% to ensure that all dogs were at a similar anesthetic plane during CT examination.12 While anesthetized, each dog received a crystalloid fluid solution (10 mL/kg/h, IV).
Computed tomography was performed with a dual-slice helical CT machine.a Transverse thoracic images were acquired at 3-mm collimation with a 50% reconstruction increment. The technical CT parameters used were the same for all examinations and included 130 kVp, 38 to 48 mAs (effective mAs; regulated by an automatic exposure control function), 0.8-second tube rotation time, and 1.85 pitch. A high-frequency reconstruction kernel was used, and images were acquired in a lung window (window level, −600 HU; window width, 1,200 HU) with a field of view of 300 × 300 mm and matrix size of 1,024 × 1,024 mm.
Immediately after anesthesia was induced, the dog was positioned in sternal recumbency, and a full helical thoracic CT scan (Si) was performed in a caudocranial direction by use of an inspiratory manual breath-hold technique (breath-hold pressure was maintained at 15 cm H2O for the entire duration of the CT scan as described7 by use of a pressure gauge mounted on the anesthetic machine) to minimize respiratory motion. Then, the dog was positioned in the designated lateral recumbency (phase 1) and allowed to lie in a natural position without the use of any positioning devices. Thoracic CT images were obtained at 3, 8, 13, 20, and 30 minutes after repositioning. Those intervals were arbitrarily chosen on the basis of findings reported in the human and canine literature,6,8,9 which suggest that atelectasis generally develops within 5 to 7 minutes (up to 15 minutes) after the patient is positioned in lateral recumbency under varying anesthetic conditions but did not progress 20 minutes after repositioning. Typically, during each scan, 10 slices were acquired perpendicular to the sternum at 5 anatomic lung sites (designated A to E; Figure 1) that were preselected to include representative areas of all lung lobes (slices A and B included representative sections of the right cranial lobe and pars cranialis of the left cranial lobe, slice C included representative sections of the pars caudalis of the left cranial lobe and the right middle lobe, and slices D and E included representative sections of the right and left caudal lobes; slice D also included a representative section of the accessory lobe). Those 5 sites were rapidly and easily determined from a topogram scan. Because of the incomplete fissure that divides the left cranial lobe into 2 parts,13 the pars cranialis and pars caudalis portions of that lobe were considered separate lobes for this study. After the last scan in lateral recumbency was completed, the dog was again positioned in sternal recumbency (phase 2), and a full helical thoracic scan was repeated (Se). That scan served as a midstudy overview of the lungs to determine whether changes had developed in sections of the lungs that were not included in the 5 selected slices. Thoracic images were then obtained at the 5 preselected sites at 5, 10, and 20 minutes after the dog was repositioned in sternal recumbency. Those time intervals were again arbitrarily chosen on the basis of reports in the literature8,9 that suggest that atelectasis either resolves within 7 minutes after the patient is repositioned in sternal recumbency or persists despite changes in body positioning. Following completion of the final scan, the dog was allowed to recover from anesthesia. After a 2-week interval, each dog underwent the same CT protocol except it was positioned in the opposite lateral recumbency during phase 1.
Representative CT images of a healthy adult spayed female Beagle obtained in the transverse (A through E) or sagittal (F) plane provided as examples of images obtained at each of 5 preselected anatomic lung sites at predetermined times before, during, and after the dog was positioned in lateral recumbency to characterize CT changes associated with atelectasis. Panels A through E were obtained at the level demarcated by boxes A through E on panel F. At each predetermined time, approximately 10 slices were acquired perpendicular to the sternum at each of the 5 sites that were preselected to include representative areas of all lung lobes. Slices A and B included representative sections of the right cranial lobe and pars cranialis of the left cranial lobe, slice C included representative sections of the pars caudalis of the left cranial lobe and the right middle lobe, and slices D and E included representative sections of the right and left caudal lobes; slice D also included a representative section of the accessory lobe. The transverse images were obtained in a lung window (window level, −600 HU; window width, 1,200 HU), and the sagittal image was obtained in a bone window (window level, 450 HU; window width, 1,500 HU). All images were acquired with the dog in sternal recumbency. For all transverse images, the left side of the image corresponds to the left side of the dog. In the sagittal image, vertebrae C7 and T13 are identified to assist with orientation.
Citation: American Journal of Veterinary Research 77, 10; 10.2460/ajvr.77.10.1082
CT image analysis
Computed tomography images were visually assessed. Measurements were performed on transverse images only and were measured by use of the primary CT computer and its dedicated softwareb by 1 investigator (CLR). All lung measurements were performed in a lung window, and images were aligned by the use of multiplanar reconstruction such that the thorax was symmetrically visualized. All lung lobes were subjectively evaluated for visible evidence of abnormally increased attenuation at each time point. To reduce bias, subjective assessment and quantitative measurement of lung tissue attenuation were performed 2 to 3 days apart instead of at the same time.
Each lung lobe was objectively measured from its paraspinal dorsalmost aspect to its most ventral tip. That measurement was divided into 3 equal horizontally orientated compartments that represented the dorsal, middle, and ventral lung fields. This method was chosen specifically for this study to evaluate the behavior of a dorsoventral lung gradient. For the most representative slice of each lung lobe obtained at each of the 5 preselected locations at each predetermined time (ie, of the 10 acquired slices at each anatomic site, the one that most closely resembled the corresponding site in Figure 1 was selected for evaluation), an ROI was drawn by use of a freehand method around the entire lung lobe of interest as well as each of its 3 compartments (dorsal, middle, and ventral; Figure 2) so that the mean attenuation and cross-sectional area for each region could be quantitatively determined. The outer lung boundary was drawn immediately inside the rib margins, and the inner lung boundary was drawn along the mediastinal organs and structures1; thus, the large blood vessels and bronchi were excluded from the ROIs. We believed that method would provide an adequate overview of the lungs in addition to minimizing radiation exposure for the patient and machine loading. Atelectasis was objectively defined on the basis of an HU scale developed for human patients that has been successfully adapted for clinical use in dogs.1 Briefly, the attenuation for normally aerated lung tissue ranged from −900 to −501 HU,1,14 with −713 HU and −846 HU considered normal for lung tissue during expiration and inspiration, respectively15; the attenuation for atelectatic lung tissue ranged from −100 to 100 HU, and that for poorly aerated lung tissue ranged from −500 to −101 HU.1,14
Representative CT image obtained in the transverse plane at site B described for the dog of Figure 1. The demarcated lung tissue is the pars cranialis of the left cranial lung lobe. Each lung lobe was objectively measured from its paraspinal dorsalmost aspect to its most ventral tip, and that measurement was divided into 3 equal horizontally oriented compartments that represented the dorsal, middle, and ventral lung fields (horizontal dashed lines). The outer lung boundary (solid white line) was drawn immediately inside the rib margins, and the inner lung boundary (dashed line) was drawn along the mediastinal organs and structures, careful to exclude the large blood vessels and bronchi. All ROIs were drawn by use of a freehand method, and the quantitative attenuation and cross-sectional area were determined for each ROI. See Figure 1 for remainder of key.
Citation: American Journal of Veterinary Research 77, 10; 10.2460/ajvr.77.10.1082
For dogs with visible changes in lung tissue attenuation, the time interval at which those changes first became evident, the lobes that were most frequently affected, and the time interval at which those changes had completely resolved were determined. On images with visible lung tissue attenuation, a small ROI (0.1 to 0.3 cm2) was manually drawn in the most severely affected area, with care taken to exclude all peripheral blood vessels. For each lung lobe with subjectively visible attenuation, the objective measurements of attenuation and cross-sectional area over time were compared with those measured for the Si scan (baseline) to determine the time interval with the highest attenuation and smallest cross-sectional area (indicative of the least aeration) and to identify which third (dorsal, middle, ventral) of the affected lobe contributed most to the observed changes. Those measurements were performed for descriptive purposes only.
Data analysis
Statistical analyses were not performed for subjective visible changes in lung tissue attenuation because the number of those changes was small. The accessory lung lobe, despite being considered part of the right lung,13 had similar CT characteristics regardless of recumbency (most likely owing to its position near midline); therefore, it was not included in any of the analyses. For each dog in each lateral recumbency (RLR and LLR), the quantitative measurements obtained for all left lung lobes were combined, as were the quantitative measurements obtained from all right lung lobes. The median lung tissue attenuation and cross-sectional area over time were plotted separately for the study population. Similar plots were also created for each specific lung lobe at each of the 5 selected locations (A through E). The resulting graphs were visually compared to determine the relationship between quantitative attenuation and cross-sectional area for the combined data and for each location.
Data were assessed for normality by assessment of descriptive statistics and visual evaluation of histograms and by means of the Anderson-Darling normality test, which was performed with statistical software.c Neither the quantitative attenuation nor cross-sectional area data were normally distributed. To normalize the attenuation data for analysis, 920 was added to each measurement so that the minimum value was 1, and then the data underwent a natural logarithm transformation. The cross-sectional area data underwent a square root transformation to normalize it for analysis. The respective quantitative changes in attenuation and cross-sectional area over time were assessed with mixed linear regression. Each model included fixed effects for phase (phase 1 [lateral recumbency] or phase 2 [sternal recumbency]), assigned lateral recumbency (RLR or LLR) during phase 1, slice location (A through E), lung lobe location (dorsal, middle, or ventral), and scan acquisition time (time) and a random effect for dog to account for repeated measures on individual dogs. Independent models were also fit for each slice–lung lobe combination. The Bonferroni correction was used for all post hoc comparisons. Similar regression models were also evaluated for each phase. All regression analyses were performed with commercially available software,d and values of P ≤ 0.05 were considered significant.
Results
Visual assessment
Visually detectable increases in lung tissue attenuation were infrequent. When dogs were scanned in LLR, an increase in attenuation was identified in left lung lobes of 4 dogs, specifically the pars cranialis portion of the left cranial lung lobe at location B for 2 dogs, the pars caudalis portion of the left cranial lung lobe at location C for 4 dogs (Figure 3), and the left caudal lung lobe at location D for 1 dog. When dogs were scanned in RLR, an increase in attenuation was detected for only 1 dog in the right caudal lung lobe at location E.
Representative CT image obtained in the transverse plane at site C described in Figure 1 for a dog that had an area of abnormally increased attenuation (arrow) consistent with a decrease in aeration in the ventral third of the pars caudalis of the left cranial lung lobe. This image was obtained 30 minutes after the dog was positioned in LFR, and the attenuation was still present in the image obtained 20 minutes after the dog was repositioned in sternal recumbency. See Figure 1 for remainder of key.
Citation: American Journal of Veterinary Research 77, 10; 10.2460/ajvr.77.10.1082
When affected, the pars cranialis of the left cranial lung lobe was generally the first lung lobe to develop abnormally increased attenuation, and that change became visually evident at 3 (n = 1 dog) and 8 (1 dog) minutes after the affected dogs were positioned in LLR. Abnormally increased attenuation of the pars caudalis of the left cranial lung lobe typically became visually evident between 8 (n = 3 dogs) and 13 (1 dog) minutes after the dogs were positioned in LLR. For the dog that developed abnormally increased attenuation in the left caudal lung lobe, that change first became evident at 13 minutes after the dog was positioned in LLR. Visible changes in attenuation most frequently involved the ventral third of the lung lobe of interest, and the pars cranialis of the left cranial lung lobe appeared to be more extensively affected than the other lobes. When evident, only mild increases in lung tissue attenuation were perceived up to 30 minutes after dogs were positioned in LLR. Moreover, those changes only partially resolved when the dogs were subsequently positioned in sternal recumbency, even after 20 minutes.
When positioned in RLR, only 1 dog developed visible evidence of abnormally increased attenuation, and that change was observed only in the right caudal lung lobe. That dog was inadvertently fed within 8 hours before induction of anesthesia, and the increase in attenuation was attributed to the dog's having a full stomach. The increase in attenuation was patchy and mild and affected the ventral third of the lobe along the subpleural margin adjacent to the thoracic wall. The abnormal attenuation was initially detected at 8 minutes after the dog was positioned in RLR; it had only slightly progressed at 30 minutes after the dog was positioned in RLR and remained fairly unchanged up to 20 minutes after the dog was repositioned in sternal recumbency. No other changes were observed in the right lung lobes (Figure 4).
Thoracic CT image obtained in the transverse plane at site C described in Figure 1 for a dog when scanned 30 minutes after it was positioned in RLR. Notice that there is no visible difference in attenuation between the right (dependent) and left (nondependent) lung fields despite the fact that this image was obtained at the level of the right middle lung lobe, which is the lobe most prone to collapse and atelectasis. See Figure 1 for remainder of key.
Citation: American Journal of Veterinary Research 77, 10; 10.2460/ajvr.77.10.1082
Quantitative evaluation of attenuation and cross-sectional area of visibly affected lobes
Quantitative measurement of attenuation and cross-sectional area of the right caudal lung lobe for the dog that developed subjectively increased attenuation while in RLR were not measured because the altered attenuation was attributed to a full stomach. For the 2 dogs with subjectively increased attenuation of the pars cranialis portion of the left cranial lung lobe, the most marked increase in quantitative attenuation was measured at 13 and 30 minutes, respectively, after the dogs were positioned in LLR. For the 4 dogs with subjectively increased attenuation of the pars caudalis portion of the left cranial lung lobe, the most marked increase in quantitative attenuation was measured at 13 (n = 1 dog), 20 (1 dog), and 30 (2 dogs) minutes after the dogs were positioned in LLR. The most marked increase in quantitative attenuation was measured at 30 minutes after positioning in LLR for the dog that developed subjectively increased attenuation of the left caudal lung lobe. The highest quantitative attenuation recorded for the study (–391 HU) was measured in the ventral third of the pars caudalis of the left cranial lung lobe of 1 dog at 30 minutes after it was positioned in LLR. The highest quantitative attenuation recorded for any small ROI manually drawn within an area with subjectively increased attenuation was −289 HU and was measured in the ventral third of the pars caudalis of the left cranial lung lobe of the same dog that had the highest quantitative attenuation recorded in the study.
Relative to baseline (Si), the decrease in cross-sectional area for lung lobes that developed abnormally increased attenuation while dogs were positioned in lateral recumbency ranged from 20% to 53%. For each dog, the smallest cross-sectional area recorded corresponded with the highest quantitative attenuation value recorded. The greatest decrease in cross-sectional area most frequently occurred in the ventral third of the affected lobe (pars cranialis of the left cranial lung lobe [n = 2 dogs] and pars caudalis of the left cranial lung lobe [3 dogs]) followed by the dorsal third of the lobe (pars caudalis of the left cranial lung lobe [1 dog] and left caudal lung lobe [1 dog]).
At the last scan of phase 2 (20 minutes after the dog was repositioned in sternal recumbency), the quantitative attenuation values (median, −669 HU; range, −700 to −618 HU) for 5 of the 7 lung lobes with subjective attenuation were higher than the corresponding baseline values, which indicated that those lobes remained poorly aerated. Further statistical analyses were not performed on those data.
Statistical descriptions
The baseline quantitative attenuation for all lung lobes at all 5 preselected anatomic sites was within the reference interval and ranged from −722 to −877 HU. When the data for all lung lobes were pooled by side (right or left) for analysis and dogs were positioned in RLR, the attenuation for the nondependent left lung lobes decreased from baseline, and the cross-sectional area for those lobes increased from baseline, and both of those values peaked after the dogs were repositioned in sternal recumbency for phase 2 (Figure 5). Meanwhile, the attenuation for the dependent right lung lobes increased from baseline, and the cross-sectional area for those lobes decreased from baseline, and both of those values again peaked after the dogs were repositioned in sternal recumbency. The findings for the nondependent and dependent lung lobes were similar when dogs were positioned in LLR. Comparison of the graphs for quantitative attenuation and cross-sectional area revealed an apparent negative association between the 2 measures (ie, when attenuation increased, there was a concurrent decrease in cross-sectional area and vice versa).
Mean quantitative attenuation (A) and cross-sectional area (B) over time as determined from CT images for the right and left lung lobes of 6 healthy adult spayed female Beagles immediately after anesthesia (baseline; Si); at 3 (L3), 8 (L8), 13 (L13), 20 (L20), and 30 (L30) minutes after the dog was positioned in RLR or LLR (phase 1); and immediately (Se) and at 5 (S5), 10 (S10), and 20 (S20) minutes after the dog was positioned in sternal recumbency (phase 2). The vertical line at L30 represents the end of phase 1 and the repositioning of dogs into sternal recumbency. In panel A, the attenuation was greatest (HU was highest, which was indicative of a decrease in aeration) for the dependent lobes and lowest for the nondependent lobes during phase 1, and the attenuation returned toward, but never fully attained, baseline values during phase 2. In panel B, the pattern for the changes in cross-sectional area for the dependent and nondependent lobes was opposite of that observed for attenuation (ie, the cross-sectional area was greatest for nondependent lobes and lowest for the dependent lobes during phase 1).
Citation: American Journal of Veterinary Research 77, 10; 10.2460/ajvr.77.10.1082
Results of the mixed linear regression analysis revealed that quantitative attenuation was not significantly (P = 0.143) associated with group (initial recumbency side) but was significantly (P < 0.001 for all) associated with time, recumbency, slice, lung lobe, and lung portion (Table 1). Similarly, cross-sectional area was not significantly (P = 0.276) associated with group but was significantly associated with time (P = 0.025), recumbency (P = 0.005), slice (P < 0.001), lung lobe (P < 0.001), and lung portion (P < 0.001; Table 2).
Results of mixed linear regression analysis to assess the respective associations of various variables with lung tissue attenuation as determined by CT examination for 6 healthy adult spayed female Beagles.
| Variable | Level | Estimate | 95% confidence interval | P value |
|---|---|---|---|---|
| Group | 1 | Referent | — | — |
| 2 | −0.192 | −0.485 to 0.101 | 0.143 | |
| Time | — | — | — | < 0.001 |
| Baseline | −0.092 | −0.143 to −0.041 | < 0.001 | |
| Lateral 3 min | −0.105 | −0.177 to −0.032 | 0.005 | |
| Lateral 8 min | −0.093 | −0.166 to −0.021 | 0.012 | |
| Lateral 13 min | −0.075 | −0.148 to −0.003 | 0.042 | |
| Lateral 20 min | −0.036 | −0.109 to 0.036 | 0.327 | |
| Lateral 30 min | −0.030 | −0.081 to 0.021 | 0.249 | |
| Sternal 5 min | −0.013 | −0.064 to 0.038 | 0.617 | |
| Sternal 10 min | −0.019 | −0.070 to 0.033 | 0.479 | |
| Sternal 20 min | Referent | — | — | |
| Recumbency | — | — | — | < 0.001 |
| Left | 0.037 | −0.017 to 0.091 | 0.179 | |
| Right | −0.050 | −0.104 to 0.004 | 0.068 | |
| Sternal | Referent | — | — | |
| Slice | — | — | < 0.001 | |
| A | −0.224 | −0.271 to −0.177 | < 0.001 | |
| B | 0.058 | 0.012 to 0.105 | 0.014 | |
| C | −0.204 | −0.254 to −0.154 | < 0.001 | |
| D | −0.042 | −0.080 to −0.004 | 0.031 | |
| E | Referent | — | — | |
| Lung lobe* | — | — | — | < 0.001 |
| Accessory | 0.482 | 0.431 to 0.532 | < 0.001 | |
| Left caudal | 0.006 | −0.032 to 0.044 | 0.769 | |
| Left cranial (pars caudalis) | 0.317 | 0.263 to 0.370 | < 0.001 | |
| Left cranial (pars cranialis) | 0.073 | 0.035 to 0.III | < 0.001 | |
| Right caudal | Referent | — | — | |
| Right cranial | Referent | — | — | |
| Right middle | Referent | — | — | |
| Lung portion | — | — | — | < 0.001 |
| Dorsal | −0.029 | −0.057 to −0.001 | 0.044 | |
| Middle | 0.045 | 0.017 to 0.073 | 0.002 | |
| Ventral | Referent | — | — |
Each dog underwent 2 CT examinations with a 2-week interval between examinations. Prior to study initiation, dogs were randomly assigned to 2 groups; dogs in group 1 (n = 3) were positioned in RLR for the first CT examination and LLR for the second examination, whereas dogs in group 2 (3) were positioned in LLR for the first examination and RLR for the second examination. Once anesthetized, each dog was positioned in sternal recumbency, and a helical transverse CT scan was acquired by use of a breath-hold technique (baseline). The dog was positioned in lateral recumbency for 30 minutes, and thoracic CT images were obtained at 5 preselected sites at 3, 8, 13, 20, and 30 minutes after repositioning (phase 1). Then, the dog was repositioned in sternal recumbency, and CT images were obtained at the 5 preselected sites at 5, 10, and 20 minutes after repositioning (phase 2). The protocol for the second examination was the same as the first except the dog was positioned in the opposite lateral recumbency during phase 1. At each time, approximately 10 slices were acquired perpendicular to the sternum at each of 5 sites that were preselected to include representative areas of all lung lobes. Slices A and B included representative sections of the right cranial lobe and pars cranialis of the left cranial lobe, slice C included representative sections of the pars caudalis of the left cranial lobe and the right middle lobe, and slices D and E included representative sections of the right and left caudal lobes; slice D also included a representative section of the accessory lobe. To normalize the attenuation data for analysis, 920 was added to each measurement so that the minimum value was 1, and then the data underwent a natural logarithm transformation.
More than 1 reference lobe was required because not all lung lobes were present in all slices, and redundant variables were removed from the model.
— = Not applicable.
Results of mixed linear regression analysis to assess the respective associations of various variables with the cross-sectional area of the lungs determined by CT examination for the dogs of Table 1.
| Variable | Level | Estimate | 95% confidence interval | P value |
|---|---|---|---|---|
| Group | 1 | Referent | — | — |
| 2 | 0.063 | −0.075 to 0.201 | 0.276 | |
| Time | — | 0.025 | ||
| Baseline | 0.110 | 0.041 to 0.178 | 0.002 | |
| Lateral 3 min | 0.125 | 0.028 to 0.221 | 0.011 | |
| Lateral 8 min | 0.III | 0.015 to 0.208 | 0.024 | |
| Lateral 13 min | 0.091 | −0.005 to 0.187 | 0.065 | |
| Lateral 20 min | 0.080 | −0.017 to 0.176 | 0.105 | |
| Lateral 30 min | 0.079 | 0.010 to 0.147 | 0.024 | |
| Sternal 5 min | 0.011 | −0.057 to 0.080 | 0.742 | |
| Sternal 10 min | 0.022 | −0.046 to 0.090 | 0.530 | |
| Sternal 20 min | Referent | — | — | |
| Recumbency | — | — | — | 0.005 |
| Left | −0.100 | −0.172 to −0.029 | 0.006 | |
| Right | −0.046 | −0.118 to 0.025 | 0.207 | |
| Sternal | Referent | — | — | |
| Slice | — | — | — | < 0.001 |
| A | -I.252 | −1.314 to −1.190 | < 0.001 | |
| B | −0.606 | −0.668 to −0.544 | < 0.001 | |
| C | −0.254 | −0.321 to −0.187 | < 0.001 | |
| D | 0.477 | 0.426 to 0.527 | < 0.001 | |
| E | Referent | — | — | |
| Lung lobe* | — | — | — | < 0.001 |
| Accessory | −0.720 | −0.787 to −0.653 | < 0.001 | |
| Left caudal | −0.189 | −0.240 to −0.139 | < 0.001 | |
| Left cranial (pars caudalis) | −0.244 | −0.315 to −0.172 | < 0.001 | |
| Left cranial (pars cranialis) | −0.366 | −0.416 to −0.315 | < 0.001 | |
| Right caudal | Referent | — | — | |
| Right cranial | Referent | — | — | |
| Right middle | Referent | — | — | |
| Lung portion | — | — | — | < 0.001 |
| Dorsal | 0.314 | 0.277 to 0.352 | < 0.001 | |
| Middle | 0.625 | 0.587 to 0.662 | < 0.001 | |
| Ventral | Referent | — | — |
The cross-sectional area data underwent a square root transformation to normalize it for analysis.
See Table 1 for remainder of key.
During phase 1 (when dogs were in lateral recumbency), attenuation was not significantly (P = 0.112) associated with group but was significantly associated with time (P = 0.034), recumbency side (P < 0.001), and lung lobe portion (P = 0.001). Cross-sectional area was not significantly associated with group (P = 0.099) or time (P = 0.614) but was significantly associated with recumbency side (P = 0.014) and lung lobe portion (P < 0.001).
During phase 2 (when dogs were in sternal recumbency following phase 1), the extent of attenuation and cross-sectional area fluctuated substantially and generally returned toward, but never fully attained, their initial baseline values. For all lung lobes, attenuation was significantly associated with lung lobe portion (P < 0.001) but not time (P = 0.479). Likewise, the cross-sectional area was significantly associated with lung lobe portion (P < 0.001) but not time (P = 0.095).
Regardless of whether a dog was positioned in lateral or sternal recumbency, the ventral third of the affected lung lobes had the highest quantitative attenuation, whereas the dorsal third had the most negative attenuation and the attenuation for the middle third fell between that for the ventral and dorsal thirds. Also, the middle third of the affected lung lobes had the largest cross-sectional area, followed by the dorsal and then ventral thirds.
Discussion
In the present study, none of the lung lobes with altered CT findings had quantitative attenuation values that were consistent with atelectasis (–100 to 100 HU), even when the small circular ROI drawn in the most visibly affected region was assessed. Thus, the dogs of this study developed only a moderate to marked decrease in lung aeration while anesthetized and positioned in lateral recumbency. For all dogs, the quantitative attenuation at baseline for all lung lobes evaluated was consistent with that described for clinically normal lungs.15
A decrease in aeration was not detected in the right lung lobes of any dog except the one that was inadvertently fed within 8 hours before anesthesia and positioned in RLR during phase 1. The peripheral changes observed in the right caudal lung lobe of that dog were attributed to a full stomach, which resulted in the pyloric antrum being gravitationally pulled to the right side when the dog was positioned in RLR, which in turn exacerbated the cranial movement of the right crus and compressed the dependent right caudal lung lobe. That mechanism was purely speculative because comparative CT images with the dog positioned in LLR were not obtained to evaluate the effect of a full stomach fundus on the dependent left lung lobes. However, no other credible causes for that finding were identified during image analysis. Interestingly, the right middle lung lobe of that dog or any other dogs in the study never developed visible attenuation changes (Figure 4) despite its being the lobe most prone to any sort of collapse, including atelectasis.16,17 It was speculated that there was less displacement of the heart when dogs were positioned in RLR than when positioned in LLR; therefore, the right middle lung lobe was not particularly compressed, even with the dogs positioned in RLR. This was observed in all 6 dogs regardless of whether attenuation changes developed in the lungs. It is also possible that the presence of the prominent cardiac notch in dogs that exposes the right side of the heart to the thoracic wall13 resulted in the heart causing less compression of the right lung lobes when the dogs were positioned in RLR, compared with that of the left lung lobes when dogs were positioned in LLR. The lack of dependent atelectasis as determined by CT imaging when the dogs of the present study were positioned in RLR was contrary to findings of another study6 in which radiographic changes consistent with dependent atelectasis were identified when dogs were positioned in RLR. The investigators of that study6 remarked that the volume of the right lung lobes was greater than that of the left lung lobes; therefore, atelectasis would be more easily perceived in the right lung lobes than in the left lung lobes. A limitation of that study6 was that the degree of inspiration at the time the radiographs were obtained was not standardized (ie, the dogs were not intubated, so a breath-hold technique could not be used). The investigators reported that many of the radiographs were obtained during expiration, which could have hindered interpretation.6
For 4 of the 7 CT examinations that resulted in visibly detectable attenuation, changes associated with a decrease in aeration of the dependent lung lobes first became visible within 3 to 8 minutes after the dog was positioned in lateral recumbency, which was consistent with the findings of other studies that involved humans (5 minutes)8 and dogs (7 minutes).6,9 For all affected lobes, the extent of impaired aeration progressed very little during the 30 minutes that the dogs remained in lateral recumbency and was not completely resolved at 20 minutes after the dogs were repositioned in sternal recumbency. Thus, in the present study, the lungs failed to return to the baseline (Si) aeration within 10 minutes as expected on the basis of results of previous studies.8,9 That failure was possibly a reflection of differences in study design and the composition of the air the dogs were breathing. Oxygen supplementation impairs lung aeration.1 In other studies in which the extent of lung aeration in dogs was assessed by CT9 or radiographic6 examination, the fairly rapid resolution of atelectasis was most likely a result of those dogs’ breathing room air. The dogs of the present study were supplemented with oxygen for the duration of the examination; thus, a degree of ongoing resorption hypoattenuation was expected.2,4,16
Results of the present study indicated that, compared with baseline measures, the dependent lung lobes had an increase in attenuation and a concomitant decrease in cross-sectional area, which were consistent with a decrease in lung aeration. Conversely, the nondependent lung lobes had a decrease in attenuation and increase in cross-sectional area, which were consistent with an increase in lung aeration. Some might question our decision to evaluate changes in the cross-sectional area of specific lung lobes in preference to the calculation of total lung volume because the change in cross-sectional area was determined for only 1 slice (location) of a lobe and did not reflect changes that might have occurred in other parts of the lobe. We felt that serial measurement of the cross-sectional area at the same location of specific lung lobes would provide insight into alterations that developed as a result of changes in recumbency. As expected, for all dogs at any point, the smallest cross-sectional area had the highest attenuation. Regardless of the lobe, the ventral third was typically the greatest contributor to the change in cross-sectional area measurement and, as such, generally had the highest attenuation of all 3 portions (ventral, middle, and dorsal). The reason the dorsal third was the greatest contributor to the change in cross-sectional area measurement in some instances was not elucidated in this study. The use of the breath-hold technique might have caused reinflation of collapsed alveoli,1 which may have affected image interpretation. However, the breath-hold technique is routinely used for thoracic CT examinations at our institution and many others. Thus, its use in the present study was clinically relevant.
Limitations of the present study included the small number of dogs evaluated and the inability to estimate intraobserver and interobserver repeatability because measurements were performed only once owing to the long time it took to capture and interpret most of the data. Evaluation of 1 slice of each lung lobe might also be considered a limitation. However, we believe that evaluation of representative slices of the various lung lobes and the method used to demarcate the ROIs provided an accurate representation of changes in lung attenuation subsequent to the development of impaired aeration over time while minimizing the radiation exposure for the study dogs. Evaluation of the entire lung at each interval would have provided the most accurate data, particularly if automated lung segmentation techniques18 were used to rapidly determine attenuation measurements. However, repeated full-thoracic scans would have ethical implications. The findings of the present study should not be extrapolated to dogs with extreme thoracic conformations (eg, Bulldogs) or pulmonary disease because the lungs of those types of dogs are likely to behave differently than the lungs of healthy Beagles. Also, the present study was not designed to investigate the visibility of subtle pathological changes (eg, small metastatic nodules) within hypoinflated regions of the lungs, and this requires further investigation.
Results of the present study indicated that healthy medium-sized dogs with normal body condition anesthetized and maintained in lateral recumbency did not develop atelectasis or lung hypoinflation sufficient enough to hamper pulmonary evaluation on CT images. Patchy areas of abnormally increased attenuation were infrequently detected in the pars cranialis and pars caudalis of the left cranial lung lobe when dogs were positioned in LLR, and those areas did not completely resolve within 20 minutes after the dogs were repositioned in sternal recumbency. Consequently, when there is a question about the presence of pathological lesions in areas of attenuation, ventilatory strategies should be applied to the lungs to resolve atelectasis. Also, because visible changes in attenuation were detected only in left lung lobes when dogs were scanned in LLR, we recommend that dogs be positioned in RLR during CT or radiographic evaluation of the lungs if they need to be imaged in lateral recumbency.
Acknowledgments
This manuscript represents a portion of a dissertation submitted by Dr. le Roux to the University of Pretoria as partial fulfillment of the requirements for the MMedVet degree.
Supported by the Department of Companion Animal Clinical Studies, Faculty of Veterinary Science, University of Pretoria.
Presented as a poster at the Annual Meeting of the American College of Veterinary Radiology, Minneapolis, October 2015.
The authors thank Dr. Melanie McLean and Dr. Zandri Whitehead for technical assistance.
ABBREVIATIONS
| HU | Hounsfield units |
| LLR | Left lateral recumbency |
| RLR | Right lateral recumbency |
| ROI | Region of interest |
Footnotes
Somatom Emotion Duo, Siemens Healthcare, Forchheim, Germany.
Somaris/5 VB20A (syngo CT 2006A), Siemens Healthcare, Berlin, Germany.
MINITAB, version 13.32, Minitab Inc, State College, Penn.
SPSS, version 21, IBM Corp, Armonk, NY.
References
1. Staffieri F, Franchini D, Carella GL, et al. Computed tomographic analysis of the effects of two inspired oxygen concentrations on pulmonary aeration in anesthetized and mechanically ventilated dogs. Am J Vet Res 2007;68: 925–931.
2. Duggan M, Kavanagh BP. Pulmonary atelectasis: a pathogenic perioperative entity. Anesthesiology 2005;102: 838–854.
3. Brooks-Brunn JA. Postoperative atelectasis and pneumonia. Heart Lung 1995;24: 94–115.
4. Hedenstierna G, Edmark L. Mechanisms of atelectasis in the perioperative period. Best Pract Res Clin Anaesthesiol 2010;24: 157–169.
5. Rothen HU, Sporre B, Engberg G, et al. Prevention of atelectasis during general anaesthesia. Lancet 1995;345: 1387–1391.
6. Barletta M, Almondia D, Williams J, et al. Radiographic evaluation of positional atelectasis in sedated dogs breathing room air versus 100% oxygen. Can Vet J 2014;55: 985–991.
7. Morandi F, Mattoon JS, Lakritz J, et al. Correlation of helical and incremental high-resolution thin-section computed tomographic imaging with histomorphometric quantitative evaluation of lungs in dogs. Am J Vet Res 2003;64: 935–944.
8. Brismar B, Hedenstierna G, Lundquist H, et al. Pulmonary densities during anesthesia with muscular relaxation—a proposal of atelectasis. Anesthesiology 1985;62: 422–428.
9. Ahlberg NE, Hoppe F, Kelter U, et al. A computed tomographic study of volume and x-ray attenuation of the lungs of Beagles in various body positions. Vet Radiol Ultrasound 1985;26: 43–47.
10. Rusca M, Proietti S, Schnyder P, et al. Prevention of atelectasis formation during induction of general anesthesia. Anesth Analg 2003;97: 1835–1839.
11. Lundquist H, Hedenstierna G, Ringertz H. Barbiturate anaesthesia does not cause pulmonary densities in dogs: a study using computerized axial tomography. Acta Anaesthesiol Scand 1988;32: 162–165.
12. Hall L, Clarke K, Trim C. Patient monitoring and clinical measurement. In: Hall L, Clarke K, Trim C, eds. Veterinary anaesthesia. 10th ed. London: Harcourt Publishers Ltd, 2001; 35–36.
13. Evans H, de Lahunta A. The respiratory system. In: Evans H, de Lahunta A, eds. Miller's anatomy of the dog. 4th ed. St Louis: Elsevier Saunders, 2013; 338–360.
14. Staffieri F, DeMonte V, DeMarzo C, et al. Influence of abdominal surgery on pulmonary atelectasis formation in dogs. In: Pugliese A, Gaiti A, Boiti C, eds. Veterinary science: current aspects in biology, animal pathology, clinic and food hygiene. Berlin and Heidelberg, Germany: Springer-Verlag Berlin Heidelberg, 2012; 141–145.
15. Schwarz T, Johnson V. Lungs and bronchi. In: Schwarz T, Saunders J, eds. Veterinary computed tomography. Chichester, West Sussex, England: John Wiley & Sons Ltd, 2011; 261–276.
16. Mai W, O'Brien R, Scrivani P, et al. The lung parenchyma. In: Schwarz T, Johnson V, eds. BSAVA manual of canine and feline thoracic imaging. Gloucester, England: BSAVA, 2008; 283–286.
17. Suter P, Lord P. Lower airway and pulmonary parenchymal disease. In: Suter P, Lord P, eds. Thoracic radiography: a text atlas of thoracic disease of the dog and cat. Wettswil am Albis, Switzerland: Peter F. Suter, 1984; 640–650.
18. McEvoy FJ, Buelund L, Strathe AB, et al. Quantitative computed tomography evaluation of pulmonary disease. Vet Radiol Ultrasound 2009;50: 47–51.
