PURPOSE: To determine the diagnostic accuracy of chest radiography and computed tomography (CT) in patients with complications during treatment for hematologic malignancies. MATERIALS AND METHODS: CT scans were obtained 1 week or less before bronchoscopic sampling or biopsy in 48 pediatric patients (age range, 8 months to 18 years at diagnosis) undergoing treatment for leukemia, lymphoma, or myeloproliferative disease. Radiographs were obtained less than 1 week before CT. Pulmonary complications comprised fungal (n = 11), viral (n = 4), and bacterial (n = 5) pneumonias; cryptogenic organizing pneumonia ([COP] n = 4); and pulmonary tumor (n = 4). Chest radiographs and CT scans were rated independently by three radiologists who were unaware of these diagnoses. RESULTS: Satisfactory diagnostic accuracy, defined by the area under the receiver operating characteristic (ROC) curve, was noted for fungal pneumonia (radiography, ROC area = 0.82; CT, ROC area = 0.78), COP (radiography, ROC area = 0.75; CT, ROC area = 0.75), and pulmonary tumor (radiography, ROC area = 0.73; CT, ROC area = 0.83). Generalizability was good for fungal pneumonia (radiography, generalizability coefficient [GC] = 0.84; CT, GC = 0.84) and COP (radiography, GC = 0.75; CT, GC = 0.99). There was no statistically significant difference in diagnostic accuracy between radiography and CT for any of the diagnoses. CONCLUSION: Radiography and CT have satisfactory accuracies for fungal pneumonia and COP. For these conditions, CT identified more true-positive cases than did radiography.
PURPOSE: To determine the diagnostic accuracy of chest radiography and computed tomography (CT) in patients with complications during treatment for hematologic malignancies. MATERIALS AND METHODS: CT scans were obtained 1 week or less before bronchoscopic sampling or biopsy in 48 pediatric patients (age range, 8 months to 18 years at diagnosis) undergoing treatment for leukemia, lymphoma, or myeloproliferative disease. Radiographs were obtained less than 1 week before CT. Pulmonary complications comprised fungal (n = 11), viral (n = 4), and bacterial (n = 5) pneumonias; cryptogenic organizing pneumonia ([COP] n = 4); and pulmonary tumor (n = 4). Chest radiographs and CT scans were rated independently by three radiologists who were unaware of these diagnoses. RESULTS: Satisfactory diagnostic accuracy, defined by the area under the receiver operating characteristic (ROC) curve, was noted for fungal pneumonia (radiography, ROC area = 0.82; CT, ROC area = 0.78), COP (radiography, ROC area = 0.75; CT, ROC area = 0.75), and pulmonary tumor (radiography, ROC area = 0.73; CT, ROC area = 0.83). Generalizability was good for fungal pneumonia (radiography, generalizability coefficient [GC] = 0.84; CT, GC = 0.84) and COP (radiography, GC = 0.75; CT, GC = 0.99). There was no statistically significant difference in diagnostic accuracy between radiography and CT for any of the diagnoses. CONCLUSION: Radiography and CT have satisfactory accuracies for fungal pneumonia and COP. For these conditions, CT identified more true-positive cases than did radiography.