Tae Iwasawa1, Tamiko Takemura2, Koji Okudera3, Toshiyuki Gotoh4, Yuma Iwao5, Hideya Kitamura6, Tomohisa Baba6, Takashi Ogura6, Mari S Oba7. 1. Department of Radiology, Kanagawa Cardiovascular and Respiratory Center, Kanagawa, Japan. Electronic address: tae_i_md@wb3.so-net.ne.jp. 2. Department of Pathology, Japanese Red Cross Medical Center, Tokyo, Japan. 3. Department of Pathology, Graduate School of Medicine, Yokohama City University, Yokohama, Japan. 4. Graduate School of Environment and Information Sciences, Yokohama National University, Yokohama, Japan. 5. National Institutes for Quantum and Radiological Science and Technology, Chiba, Japan. 6. Department of Respiratory Medicine, Kanagawa Cardiovascular and Respiratory Center, Kanagawa, Japan. 7. Department of Biostatistics and Epidemiology, Toho University School of Medicine, Yokohama, Japan.
Abstract
PURPOSE: To compare computer-aided diagnostic results with histological findings obtained by surgical biopsy and evaluate whether subpleural lesion volumes can aid identification of idiopathic pulmonary fibrosis (IPF). MATERIALS AND METHODS: We retrospectively analyzed computed tomography (CT) images of 79 patients (43 with fibrosing nonspecific interstitial pneumonia (fNSIP) and 36 with IPF) using the Gaussian Histogram Normalized Correlation (GHNC) system. We determined the H-pattern based on honeycomb and/or fibrosis with traction bronchiectasis on CT, and measured the H-pattern volume ratio at the biopsy sites and in the subpleural area. The biopsy site CT data were compared with biopsy specimens using Spearman's correlation. H-pattern volumes in the subpleural area within 2mm under the pleura (H2) were analyzed to predict IPF diagnosis and patients prognosis. RESULTS: The H-pattern volume ratio at the biopsy sites showed significant correlation with histological honeycomb (r=0.355, p<0.001), subpleural collapse (r=0.410, p<0.001), and heterogeneity (r=0.484, p<0.001). Multivariate regression analysis, adjusting for age, sex, and CT results, revealed that the H2 was a significant independent predictor of IPF diagnosis (odds ratio: 1.073; p=0.048). H2 correlated with patients' survival after adjusting for age (p=0.003). CONCLUSION: The computer-aided H-pattern volume ratio of the subpleural area indicates subpleural abnormalities quantitatively and may help diagnose IPF.
PURPOSE: To compare computer-aided diagnostic results with histological findings obtained by surgical biopsy and evaluate whether subpleural lesion volumes can aid identification of idiopathic pulmonary fibrosis (IPF). MATERIALS AND METHODS: We retrospectively analyzed computed tomography (CT) images of 79 patients (43 with fibrosing nonspecific interstitial pneumonia (fNSIP) and 36 with IPF) using the Gaussian Histogram Normalized Correlation (GHNC) system. We determined the H-pattern based on honeycomb and/or fibrosis with traction bronchiectasis on CT, and measured the H-pattern volume ratio at the biopsy sites and in the subpleural area. The biopsy site CT data were compared with biopsy specimens using Spearman's correlation. H-pattern volumes in the subpleural area within 2mm under the pleura (H2) were analyzed to predict IPF diagnosis and patients prognosis. RESULTS: The H-pattern volume ratio at the biopsy sites showed significant correlation with histological honeycomb (r=0.355, p<0.001), subpleural collapse (r=0.410, p<0.001), and heterogeneity (r=0.484, p<0.001). Multivariate regression analysis, adjusting for age, sex, and CT results, revealed that the H2 was a significant independent predictor of IPF diagnosis (odds ratio: 1.073; p=0.048). H2 correlated with patients' survival after adjusting for age (p=0.003). CONCLUSION: The computer-aided H-pattern volume ratio of the subpleural area indicates subpleural abnormalities quantitatively and may help diagnose IPF.