In-Pyeong Hwang1, Chang Min Park, Sang Joon Park, Sang Min Lee, Holman Page McAdams, Yoon Kyung Jeon, Jin Mo Goo. 1. From the*Department of Radiology, Seoul National University College of Medicine, and Institute of Radiation Medicine, Seoul National University Medical Research Center, Seoul, Korea; †Cancer Research Institute, Seoul National University, Seoul, Korea; ‡Department of Radiology, Duke University Medical Center, Durham, NC; and §Department of Pathology, Seoul National University College of Medicine, Seoul, Korea.
Abstract
OBJECTIVE: To evaluate the differentiating potentials of computed tomography texture analysis for invasive pulmonary adenocarcinomas (IPAs) from their preinvasive lesions or minimally invasive adenocarcinomas (MIAs) manifesting as persistent pure ground-glass nodules (PGGNs) larger than 5 mm. MATERIALS AND METHODS: This institutional review board-approved retrospective study included 63 patients (23 men and 40 women) with 66 PGGNs larger than 5 mm on unenhanced computed tomography from 2005 to 2013. All PGGNs were pathologically confirmed and categorized into 2 groups [IPAs (n = 11) vs preinvasive lesions (n = 41)/MIAs (n = 14)]. Each PGGN was segmented manually, and their texture features were quantitatively extracted. To identify significant differentiating factors of IPAs from preinvasive lesions/MIAs, multivariate logistic regression and C-statistic analyses were performed. RESULTS: Between IPAs and preinvasive lesions/MIAs, nodule size, volume, mass, entropy, effective diameter, and surface area were significantly different (P < 0.05), and homogeneity and gray level co-occurrence matrix inverse difference moment showed marginal significance (P < 0.10). Subsequent multivariate analysis revealed larger nodule mass [adjusted odds ratio (OR), 11.92], higher entropy (adjusted OR, 35.12), and lower homogeneity (adjusted OR, 0.278 × 10) as independent differentiating factors of IPAs. Subgroup analysis showed that larger nodule mass, higher entropy, and lower homogeneity were also significant differentiating variables of IPAs in nodules of diameter 10 mm or larger. A multiple logistic regression model using these features showed excellent [area under the curve (AUC), 0.962] and significantly higher differentiating performance compared to nodule size (AUC, 0.712) or mass (AUC, 0.788) alone. CONCLUSION: Computed tomography texture features such as higher entropy and lower homogeneity were significant differentiating factors of IPAs presenting as PGGNs larger than 5 mm and have potentials to enhance the differentiating performance.
OBJECTIVE: To evaluate the differentiating potentials of computed tomography texture analysis for invasive pulmonary adenocarcinomas (IPAs) from their preinvasive lesions or minimally invasive adenocarcinomas (MIAs) manifesting as persistent pure ground-glass nodules (PGGNs) larger than 5 mm. MATERIALS AND METHODS: This institutional review board-approved retrospective study included 63 patients (23 men and 40 women) with 66 PGGNs larger than 5 mm on unenhanced computed tomography from 2005 to 2013. All PGGNs were pathologically confirmed and categorized into 2 groups [IPAs (n = 11) vs preinvasive lesions (n = 41)/MIAs (n = 14)]. Each PGGN was segmented manually, and their texture features were quantitatively extracted. To identify significant differentiating factors of IPAs from preinvasive lesions/MIAs, multivariate logistic regression and C-statistic analyses were performed. RESULTS: Between IPAs and preinvasive lesions/MIAs, nodule size, volume, mass, entropy, effective diameter, and surface area were significantly different (P < 0.05), and homogeneity and gray level co-occurrence matrix inverse difference moment showed marginal significance (P < 0.10). Subsequent multivariate analysis revealed larger nodule mass [adjusted odds ratio (OR), 11.92], higher entropy (adjusted OR, 35.12), and lower homogeneity (adjusted OR, 0.278 × 10) as independent differentiating factors of IPAs. Subgroup analysis showed that larger nodule mass, higher entropy, and lower homogeneity were also significant differentiating variables of IPAs in nodules of diameter 10 mm or larger. A multiple logistic regression model using these features showed excellent [area under the curve (AUC), 0.962] and significantly higher differentiating performance compared to nodule size (AUC, 0.712) or mass (AUC, 0.788) alone. CONCLUSION: Computed tomography texture features such as higher entropy and lower homogeneity were significant differentiating factors of IPAs presenting as PGGNs larger than 5 mm and have potentials to enhance the differentiating performance.
Authors: Manoj Mannil; Jakob M Burgstaller; Arjun Thanabalasingam; Sebastian Winklhofer; Michael Betz; Ulrike Held; Roman Guggenberger Journal: Skeletal Radiol Date: 2018-03-01 Impact factor: 2.199