Bovine Radiology Pdf Books
- Diagnostic Imaging of Infants and Children PDF Robert G. Wells McGraw-Hill Education / Medical The essential illustrated atlas of pediatric radiology-featuring more than 3,700 images Diagnostic Imaging of Infants and Children enables you to visualize the full range of pediatric diseases and injuries using the latest diagnostic imaging techniques.
- This Website Provides Over 12000 Free Medical Books and more for all Students and Doctors This Website the best choice for medical students during and after learning medicine. Direct Diagnosis in Radiology Head and Neck - edition pdf Head and Neck Imaging (eBook) See more.
Can identify anatomic structures accurately. Alternative imaging studies – including CT, MRI, ultrasound, GI, urogram, and other contrast media studies -allow advanced diagnostic interpretation. A companion CD includes all of the radiographic images in the book for digital monitor viewing. File type: PDF File size: 11.8 MB.
Language: English French
Abstract
Medical records of 42 cows that underwent both thoracic radiographic and postmortem examinations within a period of 7 days were reviewed to develop an evaluation grid to interpret bovine thoracic radiographs and to determine the sensitivity and the specificity of thoracic radiographs for detection of thoracic lesions, based on postmortem examination. Most cows (64%) had clinical signs of respiratory disease, whereas 19% showed signs of cardiac problems. The sensitivity and specificity of radiographs for identifying cows with thoracic lesions were 94% and 50%, respectively. In this study, with a prevalence of thoracic lesions of 86%, the positive- and negative-predictive values were 92% and 57%, respectively. This study provides an evaluation grid that allows standardization of the reading of bovine thoracic radiographs and the identification of most thoracic lesions. Bovine thoracic radiographs are useful in detecting thoracic lesions in cows.
Résumé
Détection radiographique des lésions thoraciques chez la vache adulte : étude rétrospective de 42 cas (1995–2002). Les dossiers médicaux de 42 vaches ayant été soumises à la fois à des examens radiographiques thoraciques et à une nécropsie en dedans d’une période de 7 jours ont été revus dans le but d’élaborer une grille d’évaluation. Cette grille a servi à interpréter les radiographies thoraciques et à déterminer la sensibilité et la spécificité des radiographies dans la détection des lésions thoraciques, en les comparant avec les résultats obtenus aux examens post mortem. La majorité des vaches (64 %) montraient des signes cliniques de maladie respiratoires alors que 19 % montraient des signes de problèmes cardiaques. La sensibilité et la spécificité des radiographies servant à l’identification des vaches présentant des lésions thoraciques étaient respectivement de 94 % et 50 %. Dans cette étude, avec une prévalence de lésions thoraciques de 86 %, les valeurs prédictives positives et négatives étaient respectivement de 92 % et 57 %. Cette étude fournit une grille d’évaluation qui permet la standardisation des lectures des radiographies thoraciques bovines et l’identification de la majorité des lésions thoraciques. Les radiographies thoraciques sont utiles pour détecter les lésions thoraciques chez la vache.
(Traduit par Docteur André Blouin)
Introduction
Thoracic radiography is a useful tool for investigating lower respiratory tract integrity (). It is used mostly to establish the extent and severity of disease, and it is a practical, noninvasive, ancillary method for monitoring the recovery of a patient (). Many questions remain in the interpretation of bovine thoracic radiographs and, to our knowledge, no previous study has reported the correlation between radiographic evaluation and postmortem findings in cattle. The appearance of selected respiratory conditions, later confirmed by findings on necropsy, has been described (3). Bargai et al (4) and Farrow () illustrated the various types of lung lesions observed on thoracic radiographs in cattle, but these were not confirmed by necropsy. Two retrospective studies on radiographic diagnosis of thoracic disorders in cattle have been published, but neither used correlation with necropsy results (5,). Other authors have suggested that their radiographic findings did not always correlate with histopathologic findings and that, often, lesions at postmortem examination were more severe than suspected from the radiographs (5). The objectives of this study were to develop an evaluation grid with which to interpret bovine thoracic radiographs and determine the sensitivity and the specificity of thoracic radiographs for detection of thoracic lesions, based on postmortem examination as the gold standard.
Materials and methods
Medical records of 7244 cattle older than 13 mo that were presented to the Centre Hospitalier Universitaire Vétérinaire (CHUV), Université de Montréal, between June 1995 and December 2002 were reviewed. Cattle were selected for the study if they were older than 18 mo and had undergone thoracic radiographic and postmortem examinations within a 7-day interval. Medical records were reviewed to obtain signalment, reason for presentation, respiratory clinical signs, rectal temperature, hematological profile [white blood (cell) count (WBC), neutrophil and differential counts], fibrinogen concentration, albumin/globulin ratio, and necropsy findings.
Radiographic examination
Thoracic radiography had been performed with an X-ray unit (Siemens Gigantos Optimatic 1250 X-ray; Siemens Canada, Mississauga, Ontario). Standing left lateral radiographs had been taken on regular medical X-ray film, in combination with a high-speed screen and a grid. Films had been processed in an automatic daylight processor (Kodak RP × omat; Maranda Lauzon, Laval, Quebec). Examination had included 4 standard views of the lungs, cranioventral, caudoventral, craniodorsal, and caudodorsal, with exposure settings of 80, 16, 12, and 8 mAs, respectively, and 133 kVp (Figure 1).
Illustration of the 4 partially overlapping 14- × 17- views used for detection of thoracic lesions on a standing adult cow. A — cranioventral view; B — caudoventral view; C — craniodorsal view; D — caudodorsal view. The rectangular-shape line indicates the 13th rib.
An evaluation grid was developed to identify lung lesions and other thoracic findings involving the pleural cavity, mediastinum, and pericardium (Figure 2). Roentgen signs of pulmonary parenchymal abnormalities (alveolar, interstitial, and bronchial pattern) and their distribution (focal, multifocal, regional, diffuse), location (ventral, dorsal, cranial, caudal), and severity (mild, moderate, severe) were recorded. Presence of cavitary lesions (abscess, emphysematous bulla, or both) was noted. If a consensus among the radiologists could be obtained in specifying the likelihood of whether a cavitary lesion was an abscess or an emphysematous bulla, it was recorded on the evaluation grid, otherwise only a cavitary lesion was noted. Identification of pneumothorax, pleural effusion, pericardial effusion, pneumomediastinum, mediastinal masses, pneumopericardium, thoracic masses, or foreign bodies was also recorded. All radiographs were reviewed by 3 radiologists at the CHUV who were not aware of the history and necropsy findings. The radiologists discussed the findings and a consensus was recorded.
Evaluation grid designed for systemic reading of bovine thoracic radiographs.
Postmortem examination
Data regarding pulmonary parenchymal lesions, their distribution (focal, multifocal, regional, diffuse), location (ventral, dorsal, cranial, caudal), the anatomic localization affecting 1 or more of the following regions (interstitium and/or alveoli and/or bronchi), and severity (mild, moderate, severe) were obtained from each report. The presence of a cavitary lesion (abscess, emphysema, or both), pleural or pericardial effusion, pneumomediastinum, pneumothorax, and cardiac lesions were also recorded.
Analysis
The sensitivity, specificity, and positive- and negative-predictive values (PPV and NPV) of radiographs to detect pulmonary and thoracic lesions (alveolar, interstitial, bronchial, cavitary, abscess, emphysematous bulla, pleural effusion, pneumodiastinum) were estimated. A positive predictive value (PPV) is the probability that an animal that is positive on test, actually has the disease. The negative predictive value is the probability that an animal that is negative on test does not have the disease (7). Confidence intervals (95%) were calculated for all parameters evaluated by using the binomial exact method with software [NCCS (New Control Computer System), Kaysville, Utah, USA].
Bovine Radiology Pdf Books Pdf
Results
Forty-two cattle met the inclusion criteria for the study. Signalment, selected clinical and hematological findings upon admission, and delay between radiographs and postmortem examination for the 42 cows are shown in Table 1. Breeds represented were Holstein (n = 36), Ayrshire (n = 3), Jersey (n = 1), brown Swiss (n = 1), and Limousin (n = 1). Median age at presentation was 5.1 y (mean age: 5.4 y; range: 19 mo to 12 y). Cattle had been admitted to CHUV for 1 or more of the following problems: anorexia, respiratory abnormalities, fever, lethargy, and poor milk production. Twenty-seven cows (64%) had had clinical signs of respiratory disease or abnormal findings at the physical examination, including elevated or abnormal lung sounds and tachypnea, respiratory distress, or both. Eight cows had had clinical signs of cardiac problems (19%). Thoracic radiographs had been taken to exclude pulmonary parenchymal lesions in 6 cows. The reason that thoracic radiographs were taken was not specified in 1 case. Fourteen cows (33%) had had an elevated rectal temperature (> 39.2°C). Most of the cows (79%) had had a serum albumin/globulin ratio of less than 0.84, increased fibrinogen level (> 5 g/L), or both. Twenty-eight cows (67%) had shown leukocytosis (WBC > 12 × 109/L; reference range, 4 to 12 × 109/L), neutrophilia (> 4 × 109/L; reference range, 0.6 to 4 × 109/L), or both, and 15 cows (36%) had shown a left shift or neutropenia (< 0.6 × 109/L). Most animals (33/42) had shown 2 or more parameters indicative of an inflammatory process. In 9 cases, no inflammatory process could be detected on blood analysis.
Table 1
Signalment, selected clinical and hematological findings upon admission and delay between radiographs and the postmortem examination of 42 cows
| Cow | Age (years) | Breed | Clinical signsa | T (°C) | HR (beats/min) | RR (breaths/min) | WBC (× 109/L) | Neutrophil (× 109/L) | Immat. neutro (× 109/L) | Ratio A/G | Fibrinogen (g/L) | Delayb (days) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 6.8 | H | Cardiac | 37.5 | 120 | 36 | 7.5 | 5.3 | 0.5 | 0.94 | 8 | 1 |
| 2 | 4.6 | H | Respiratory | 40.1 | 84 | 96 | 7.97 | 4.5 | 0.8 | 0.63 | 11 | 2 |
| 3 | 1.7 | H | Respiratory | 38.5 | 48 | 16 | 7.75 | 2.2 | 1.18 | 3 | 4 | |
| 4 | 2.1 | H | Respiratory | 38.1 | 100 | 28 | 25.4 | 17.7 | 1.7 | 1.01 | 15 | 2 |
| 5 | 7.7 | H | Respiratory | 39.3 | 100 | 40 | 7.5 | 1.1 | 3.4 | 0.92 | 5 | 0 |
| 6 | 8.2 | H | Respiratory | 36.6 | 120 | 48 | 12.8 | 8.3 | 0.2 | 1.17 | 8 | 2 |
| 7 | 2.2 | Ayr | Cardiac | 39 | 104 | 28 | 11.4 | 5.7 | 0.74 | 5 | 0 | |
| 8 | 3.8 | H | Respiratory | 39.2 | 72 | 60 | 5.2 | 3.6 | 1.11 | 9 | 2 | |
| 9 | > 2 | H | Abdominal | 38.5 | 104 | 28 | 8.1 | 6.1 | 1.04 | 4 | 0 | |
| 10 | 5.5 | H | Respiratory | 39.7 | 120 | 52 | 10.6 | 6.9 | 0.4 | 5 | 0 | |
| 11 | 3.4 | H | Respiratory | 38.7 | 88 | 120 | 7.2 | 2.9 | 0.88 | 7 | 2 | |
| 12 | 7.2 | H | Respiratory | 38.2 | 120 | 30 | 11.3 | 9 | 0.9 | 0.61 | 7 | 2 |
| 13 | 6.2 | H | Respiratory | NA | NA | NA | 1.1 | 0.2 | 0.86 | 6 | 2 | |
| 14 | 7.6 | Lim | Respiratory | 38.4 | 108 | 32 | 16.9 | 14.2 | NA | 6 | 6 | |
| 15 | 4.6 | H | Respiratory | 40.7 | 84 | 104 | 3.8 | 2.6 | 1.27 | 4 | 1 | |
| 16 | 4.1 | H | Respiratory | 39.3 | 100 | 52 | 5.95 | 2.3 | 0.1 | 0.99 | 6 | 3 |
| 17 | 6.1 | H | Respiratory | 39.2 | 104 | 60 | 4.5 | 1 | 0.8 | 0.72 | 9 | 3 |
| 18 | 8.7 | H | Abdominal, mastitis | 39.6 | 76 | 56 | 7.6 | 4.9 | 0.85 | 1 | 2 | |
| 19 | 9.6 | H | Mastitis | 38.6 | 84 | 36 | 8 | 5.8 | 0.59 | 3 | 2 | |
| 20 | 5.2 | H | Respiratory | 40.6 | 120 | 90 | 5.4 | 2.5 | 1.13 | 8 | 4 | |
| 21 | 3.3 | H | Cardiac | 38.2 | 108 | 32 | 14 | 10.1 | 0.62 | 3 | 2 | |
| 22 | 3.8 | Ayr | Respiratory | 39.7 | 100 | 52 | 12.5 | 6.9 | 0.55 | 7 | 4 | |
| 23 | > 2 | H | NA | 38.8 | 52 | 20 | 5.6 | 2.8 | 0.92 | 7 | 4 | |
| 24 | 2.4 | H | Respiratory | 39 | 72 | 48 | 18.5 | 14.8 | 0.4 | 0.86 | 7 | 3 |
| 25 | 1.7 | H | Respiratory | 38.1 | 68 | 28 | 11.2 | 7.2 | 0.2 | 0.5 | 12 | 1 |
| 26 | 5.3 | H | Respiratory | 40.7 | 108 | 68 | 4.2 | 2.3 | 0.04 | 0.72 | 6 | 1 |
| 27 | 3.2 | Jer | Respiratory | 37.8 | 72 | 54 | 9.3 | 2.5 | 2.10 | 1.06 | 10 | 4 |
| 28 | 3.9 | H | Respiratory | 39.2 | 82 | 16 | 11.4 | 7.8 | 1.8 | 1.22 | 6 | 5 |
| 29 | 7.4 | H | Respiratory | 39 | 108 | 26 | 4.6 | 3 | 0.84 | 4 | 4 | |
| 30 | 12 | H | Cardiac | 37.5 | 120 | 36 | 2.8 | 1.8 | 0.43 | 8 | 1 | |
| 31 | 3 | Ayr | Respiratory | 39.2 | 96 | 72 | 18.5 | 11.3 | 0.88 | 10 | 4 | |
| 32 | 7.3 | H | Respiratory | 39.2 | 72 | 44 | 10 | 6.3 | 1.01 | 1 | 2 | |
| 33 | 5.3 | H | Cardiac | 39.5 | 100 | 60 | 7.4 | 4.1 | 1.59 | 2 | 2 | |
| 34 | 8.8 | H | Abdominal | 39.5 | 104 | 40 | 17.3 | 6 | 0.99 | 5 | 3 | |
| 35 | 5 | H | Multiple masses, lameness | 39.8 | 116 | 84 | 5.5 | 4.2 | 0.9 | 2 | 1 | |
| 36 | 4 | H | Cardiac | 38.5 | 100 | 24 | 18.5 | 16.1 | 0.1 | 0.62 | 2 | 7 |
| 37 | 3.7 | Sui | Respiratory | 38.9 | 84 | 60 | 7.2 | 5.4 | 0.68 | 11 | 2 | |
| 38 | 5.2 | H | Respiratory | 39.2 | 120 | 96 | 13.4 | 11.7 | 0.91 | 9 | 1 | |
| 39 | 4 | H | Cardiac | 38.9 | 102 | 42 | 16.9 | 11.4 | 0.54 | 10 | 1 | |
| 40 | 7.3 | H | Renal | 39 | 80 | 36 | 12 | 8.1 | 0.29 | 11 | 7 | |
| 41 | 11.6 | H | Respiratory | 39.3 | 88 | 64 | 9.3 | 6.6 | 0.39 | 8 | 1 | |
| 42 | 3.1 | H | Cardiac | 38.9 | 112 | 36 | 10.3 | 5.5 | 0.49 | 4 | 2 |
The mean and median time between radiography and death was 2 d. Thirty-seven cows had thoracic lesions, based on the thoracic radiographs. Prevalence of thoracic lesion by type was as follows: interstitial disease (n = 20/42 or 48%), cavitary lesions (n = 19/42; 45%), alveolar disease (n = 15/42; 36%), pneumo-mediastinum (n = 10/42; 24%), bronchial disease (n = 9/42; 21%), and pleural effusion (n = 8/42; 19%). Twenty-eight of the 42 cows had had more than 1 type of thoracic lesion, based on radiographs. Seventeen combinations of 2 or more thoracic radiographic signs were identified. The radiographic signs most commonly observed in these combinations included cavitary lesions, interstitial pattern, and alveolar pattern.
The sensitivity and specificity of radiographs for the detection of thoracic lesions, based on postmortem examination, are shown in Table 2. At postmortem examination, 36 of the 42 (86%) cows had thoracic lesions. Three cows had false-positive radiographs; 1 was identified as having interstitial disease localized in the caudodorsal lung area and 2 other cases were suspected of having bronchial disease. All 3 were classified as having mild lesions. Two of the 42 cows had false-negative radiographs; 1 had a few atelectatic lobules and 1 showed a localized pleuritis and some discrete disseminated hemorrhagic foci. On histological examination, these lesions proved to be bacterial emboli with inflammatory cells. The sensitivity and the specificity of radiographs for identifying cows with thoracic lesions in this particular population were 94% and 50%, respectively. In this study, with a prevalence of thoracic lesions of 86%, the PPV and NPV were 92% and 57%, respectively.
Table 2
Sensitivity and specificity of radiographs for the detection of thoracic lesions compared with postmortem findings in 42 cows
| +/+ | −/− | −/+ | +/− | Total | Sensitivity (%)a | Specificity (%)a | |
|---|---|---|---|---|---|---|---|
| Thoracic lesion | 34 | 3 | 2 | 3 | 42 | 94 (87–100) | 50 (10–90) |
| Alveolar disease | 12 | 15 | 8 | 2 | 37 | 60 (39–82) | 88 (73–100) |
| Interstitial disease | 9 | 14 | 5 | 5 | 33 | 64 (39–89) | 74 (54–93) |
| Cavitary lesion | 16 | 17 | 6 | 3 | 42 | 73 (54–91) | 85 (69–100) |
| Abscess | 3 | 28 | 7 | 4 | 42 | 30 (2–58) | 88 (76–99) |
| Emphysema | 6 | 23 | 10 | 3 | 42 | 38 (14–61) | 88 (76–100) |
| Pleural effusion | 5 | 32 | 3 | 2 | 42 | 63 (29–96) | 94 (86–100) |
| Pneumothorax | 0 | 35 | 2 | 5 | 42 | 0 | 88 (77–98) |
| Pneumomediastinum | 2 | 33 | 2 | 5 | 42 | 50 (1–99) | 87 (76–98) |
+/+ = true-positive; −/− = true-negative; −/+ = false-negative; +/− = false-positive
Pulmonary parenchymal lesions had been impossible to detect because of pleural effusion in 5 cows. Of the remaining 37 cows, 12 had been identified as having an alveolar pattern on radiographs, whereas in 20 cows, alveolar disease had been detected only at postmortem examination. In all cases of alveolar disease for which an alveolar pattern had been identified on radiographs, infectious pneumonia was identified at necropsy. Of the 8 false-negative cases, 5 had been identified as having an interstitial pattern, whereas 3 had no pulmonary parenchymal disease. The sensitivity and specificity of radiographs for identifying cows with alveolar disease were 60% and 88%, respectively.
Interstitial pattern had been masked by alveolar lesions in 4 cases. Of 14 cows having an interstitial disease at postmortem examination, 9 had been correctly identified with an interstitial pattern on radiographs and 14 of 19 cows that were free of interstitial disease at necropsy had been correctly identified on radiographs. The sensitivity and the specificity of radiographs for identifying cows with interstitial disease were 64% and 74%, respectively. All true-positive cases (9/14) were identified as being severe. The postmortem diagnoses were interstitial pneumonia (n = 3), embolic pneumonia (n = 3), chronic pneumonia (n = 1), tumoral infiltration (n = 1), and diffuse emphysema (n = 1).
Of the 9 cases that had been diagnosed with bronchopneumonia at postmortem examination, 5 had been identified with an alveolar pattern, which may or may not have been associated with an interstitial pattern on radiographs. In 1 case, the lesions had possibly been masked by pericardial effusion (noted on postmortem examination), and in the other 3, the lesions had been interpreted as having an interstitial pattern. Most cases (8/9) had been identified as having a cavitary lesion. Of the 9 cows that had been identified with a bronchial pattern on radiographs, no thickening of the bronchial walls or peribronchial tissues was noted on postmortem examination. All cases had had thoracic histopathologic lesions, such as fibrinous pneumonia, cardiac and secondary pulmonary lesions, and tumoral thoracic infiltration. All levels of severity (mild, moderate, severe) had been identified in these cases. The average age of cows that had been identified with a bronchial pattern was higher than that of the study population (6.6 y for those with a bronchial pattern compared with 5.0 y for the other cows). The sensitivity and the specificity of radiographs for identifying cows with bronchial disease were not calculated because of the shortage of true positive cases. For each case, the radiographic distribution, localization, and severity of pulmonary parenchymal lesions, except for the bronchial patterns, were evaluated together, in order to have a single value for each to compare with the distribution, localization, and severity described on postmortem examination. Since it was not possible to correlate the radiographic appearance of the bronchial pattern with the postmortem findings, bronchial pattern was not included in the determination of distribution, localization, and severity. The radiographic distribution and localization of pulmonary parenchymal lesions were in accordance with postmortem findings in 49% and 51% of cases, respectively. The results of comparison between radiographic observation and postmortem findings for the severity of pulmonary parenchymal lesions tended to form 2 groups: a) minimal, or absence of, lesions on either radiographs or postmortem examination, and b) moderate or severe lesions. The severity of pulmonary parenchymal lesions on radiographs in both groups was in accordance with postmortem findings in 26 of 32 cases (81%). Five cases were excluded because of the lack of severity score on the evaluation grid or on the postmortem report.
Journal Of Chemical Education
The sensitivity and the specificity estimated for the presence of cavitary lesion were based on indications of abscess or emphysema at postmortem examination (Table 2). When a cavitary lesion had been identified on radiographs without specifying its nature (abscess or emphysematous bulla), half of the cases showed an abscess and the other half were diagnosed with an emphysematous bulla on postmortem examination. Eight of 9 cows with an abscess at necropsy had been identified as having a cavitary lesion on radiographs. The sensitivity and the specificity estimated for the presence of pneumothorax, pleural effusion, and pneumomediastinum are shown in Table 2. No cow was identified with pericardial effusion on radiographs. On postmortem examination, 8 cows had evidence of pericardial fluid, which varied from 50 mL to 5 L. In most cases (5/8), pleural effusion was also present. In 2 other cases (2/8), the amount of fluid was low (< 1 L) and had not been be detected on radiographs. In 1 case with 5 L of pericardial fluid present at postmortem examination, pleural effusion had been diagnosed on radiographs. No cow was identified as having had a mediastinal mass or a pneumopericardium. One cow had shown a thoracic mass, which was an abscess, and 1 cow had been identified as having a foreign body in the thorax, which was confirmed at necropsy.
At postmortem examination, 13 of 42 cows had cardiac disease, including vegetative endocarditis (n = 6), tumoral infiltration [lymphoma (n = 2) and mesothelioma (n = 1)], foreign body in the myocardium (n = 1), pericarditis (n = 1), cardiomyopathy (n = 1), and right cardiac failure (n = 1). Eight of 13 cows had pleural effusion, pericardial effusion, or both. Of these 8 cows, 6 had been identified on radiographs as having pleural effusion. These 6 cows represented 75% of positive cases for pleural effusion on radiographs.
Discussion
Elaboration of an evaluation grid
The evaluation grid created in this study facilitated systematic reading and identification of most thoracic lesions, except pleuritis and atelectasis, affecting lungs, pleura, mediastinum, and pericardium. Recognition of a clinically significant bronchial pattern was highly subjective.
Limits of the retrospective study
The age, sex, and breed of the studied population were comparable with those of the average population of cows presented at the CHUV during the study period. The inclusion criteria for this retrospective study required cows to have undergone thoracic radiography and a postmortem examination. These criteria obviously created a bias towards a population where thoracic lesions were very prevalent, which was confirmed by the high level of detection of thoracic lesions on the radiographs (37/42) and at the postmortem examination (36/42). The limitations of the study were as follows: first, the sample of cows, particularly healthy cows, was small, which had an influence on the estimated PPV and NPV. Consequently, these values may not reflect the status of a normal population. Second, the maximum delay between radiographs and postmortem examination was set at 7 d so as to include a larger number of cows, but significant changes could have occurred between these 2 time points. Third, the postmortem examinations were performed by different pathologists without a standardized protocol. Variability among postmortem examinations may have contributed to the omission of small lesions.
Sensitivity and specificity of radiographs
The greatest correlation between radiographs and postmortem findings occurred in moderate to severe pulmonary diseases. Mild changes reported may be over-interpreted. With a sensitivity of 94%, radiographs are useful in detecting thoracic lesions in cows. Thus, in a population with a high prevalence of respiratory and cardiac problems, the probability of detecting thoracic lesions on radiographs in affected cows is excellent. If these results were extrapolated to a population of cows in which the prevalence of respiratory and cardiac problems was 10%, radiographs with a PPV of only 21% would be less reliable for identifying cows with thoracic lesions. Radiographs, however, would be useful in identifying healthy cows without thoracic lesions, since the NPV would be 99%. In this study, radiographs were found to be less sensitive in detecting thoracic lesions in the cranioventral part than in the other parts of the lung lobes, which is understandable because of the opacity caused by the shoulder (technically inevitable). Thoracic radiographs were also found to be less sensitive in detecting small lesions like hemorrhagic foci and small adherences. The specificity of radiographs to detect the normal thorax was only 50%, based on a small sample (6/42) of normal thoraxes. The radiographic findings for the 3 false-positive cases were all classified as being minimal lesions: the lesions described were those of a bronchial pattern in 2 cases and, of an interstitial pattern in 1 case. Since, bronchial patterns are often observed in older cows, they may be an age-related phenomenon and overinterpreted. On the other hand, since these lesions were not equally distributed on the radiographs, they may not have been sampled for histopathologic examination. In the case in which a minimal interstitial pattern was observed in the area of the caudal vena cava, the area is difficult to evaluate due to superimposition of many structures.
Moderate sensitivity detection of alveolar and interstitial pattern
Possible explanations for the moderate (66%) sensitivity for the detection of an alveolar pattern on thoracic radiographs, based on postmortem findings, include the delay of 4 d between taking the radiographs and the necropsy in 4 cases. In this interval, the disease may have progressed. In the other cases, small hemorrhagic foci, edema, microabscess, or all 3, were reported as being found on postmortem examinations but not detected on radiographs. Interestingly, in cases in which bronchopneumonia was diagnosed at postmortem, either an alveolar pattern, an interstitial pattern, or both, had been identified on radiographs. They were generally associated with a cavitary lesion (8/9). Radiographs were sensitive in identifying interstitial patterns in cows having interstitial pneumonia, embolic pneumonia, and chronic pneumonia. Five cases with interstitial disease at postmortem examination were not identified as having an interstitial pattern on the evaluation grid. Again, the delay of 4 d between taking the radiographs and doing the necropsy may be the explanation. In these cases, lesions that had been found at necropsy were small and limited to the periphery of the lung lobes. Lesions had been described as atelectasis, pneumatocele, and bronchioli wall fibrosis.
Bronchial pattern
All cows identified with bronchial patterns had other thoracic lesions and an average age (6.6 y) that was higher than that of the other cows in the study population (5.0 y). This suggests that age-related changes such as mineralization of bronchial cartilage, as observed in older dogs, may occur. Such changes would increase the radiographic opacity of the bronchial wall (). This remains to be investigated.
Distribution, localization, and severity
The moderate correlation in the distribution and localization of thoracic lesions between radiographs and postmortem examination can be attributed to several factors: the delay between thoracic radiographs and postmortem examination, the lesion being obscured by bone or vascular structures, or the small size of a lesion; for example, uncalcified lesions less than 6 mm in diameter were shown to be undetectable on equine radiographs (). Radiographic identification of multiple moderate to severe lesions generally correlated with a high severity level (moderate or severe) of the lesions observed at postmortem examination. One case did not follow this rule, as minimal lesions were identified at postmortem examination compared with a severe interstitial pattern and a moderate bronchial pattern noted on the evaluation grid. This finding could be explained, in part, by the presence of a fibrous adherence between the right cranial lung lobe and the pericardium with atelectasis of the lobe. Identification of a bronchial pattern on radiographs is probably due to an age related change rather than a sign of bronchial disease. Moreover, slight underexposure can make a big difference for the interpretation of thoracic radiographs in cattle; for example, by making the main caudal lobe bronchi look overly prominent.
Detection of cavitary lesion
The sensitivity of radiographs for detection of a cavitary lesion was moderate (73%), and all false-negative cases were attributed to lesions of emphysema. Emphysema can be detected easily when large emphysematous bullae are present. Diffuse emphysema, however, may only make the dorsal part of the lungs appear more radiolucent and the thin margins of the bulla may not be easily identified on radiographs. Overexposure of the films may contribute to a false negative diagnosis. When caudal lung lobes appear more radiolucent than usual on radiographs with adequate exposure, diffuse emphysema should be suspected. Interestingly, when a cavitary lesion was found on a radiograph and could not be further classified as an abscess or emphysema, the postmortem diagnosis was abscess in 50% of cases and emphysematous bullae in 50% of cases. Radiographs were more specific (88%) than sensitive (30%) for the detection of an abscess. However, 89% of cows with an abscess found at postmortem examination were identified with either an abscess or a cavitary lesion on radiographs. This finding suggests that in most cases in which an abscess is present in the thoracic cavity, a cavitary lesion can be identified on radiographs.
Pleural and pericardial effusion
One of the difficulties encountered in this study was to distinguish between pleural and pericardial effusion and to describe cardiac lesions. The detection of a pleural effusion on radiographs was moderately sensitive (63%). This could be explained by the small amount (< 2 L) of pleural fluid detected at postmortem examination in 2 cases [in horses, it has been estimated that approximately 1.5 L of normal saline is required before a discernable fluid opacity is consistently identified on thoracic radiographs (10)]. However, in another case, the amount at postmortem examination was estimated at 10 L, a quantity sufficient to be detected on radiographs; the time gap between the radiographs and the necropsy may again be the cause. Despite the inability to specify the exact nature of the effusion, on radiographs, 75% of cases described with pleural effusion on radiographs had an underlying cardiac problem. Most of the cases (86%) identified with pleural effusion on radiographs showed pericardial effusion, with or without an associated pleural effusion, on necropsy. Consequently, when pleural effusion is identified on radiographs, a cardiac problem should be suspected. A better knowledge of the radiographic appearance of the cardiac silhouette could also contribute to the identification of an underlying cardiac problem.
In summary, this study provides an evaluation grid for a systematic reading of bovine thoracic radiographs. Thoracic radiographs are a useful tool to detect thoracic lesions in cattle, particularly when lower respiratory tract or cardiac diseases are suspected.
Acknowledgments
The authors thank Drs. Denise Bélanger, Guy Beauchamp, and John Middleton for their statistical advice, and Dr. André Desrochers for his advice and technical help for image editing. CVJ