PURPOSE: To compare the morphologic abnormalities on thin-section computed tomographic (CT) images in a group of patients with histopathologically confirmed nonspecific interstitial pneumonia (NSIP) or usual interstitial pneumonia (UIP) and a clinical presentation of idiopathic pulmonary fibrosis. MATERIALS AND METHODS: Thin-section CT imaging patterns and distribution of disease in 53 patients with histologic diagnoses of NSIP (n = 21) or UIP (n = 32) were quantified retrospectively and independently by four observers. The appearances of NSIP and UIP at CT were compared with univariate and multivariate techniques. RESULTS: The use of thin-section CT proved to have moderate sensitivity (70%), specificity (63%), and accuracy (66%) in the diagnosis of NSIP. An increased proportion of ground-glass attenuation was the cardinal feature of NSIP at CT (odds ratio: 1.04 for each 1% increase in the proportion of ground-glass attenuation). A histologic diagnosis of NSIP was most frequent (in 24 of 35 observations [69%]) when ground-glass attenuation predominated, and was more frequent with mixed (35 of 79 observations [44%]) than with predominantly reticular disease (25 of 98 [26%] observations, P < .005). Logistic regression analysis of the data indicated that misdiagnosis of UIP in patients with NSIP was associated with less ground-glass attenuation (P < .005) at CT and a subpleural disease distribution (P = .02), with the converse being true for UIP cases misdiagnosed as NSIP. CONCLUSION: In patients with a clinical presentation of idiopathic pulmonary fibrosis, the accuracy of thin-section CT in identifying NSIP is considerably higher than previously reported. At CT, NSIP is characterized by more ground-glass attenuation and a finer reticular pattern than is UIP. Nevertheless, considerable overlap in thin-section CT patterns exists between NSIP and UIP.
- Computed tomography (CT), electron beam
- Lung, CT
- Pneumonia, nonspecific interstitial and fibrosis
- Pneumonia, usual interstitial
ASJC Scopus subject areas
- Radiology Nuclear Medicine and imaging