CT texture analysis in diagnosis of invasiveness of lung adenocarcinoma manifesting as pure ground glass nodules less than 10 mm

Abstract

Objective To assess the value of CT texture analysis in diagnosing the invasiveness of lung adenocarcinoma manifesting as pure ground glass nodules less than 10 mm. Methods The preoperative CT data of 148 patients manifesting as pure ground glass nodules less than 10 mm who underwent operation were analyzed retrospectively. Among 148 cases of pure ground glass nodules, 73 were preinvasive lesions and 75 were invasive lesions of lung adenocarcinoma. ImageJ software was used to extract the texture features of the images and then the parameters that differed significantly between the preinvasive lesions and invasive lesions were identified. Spearman correlation analysis was used for feature dimension reduction, and then independent predictors were selected out. Receiver operator characteristic curve was then used to evaluate the diagnostic efficiency of independent predictors. Results Among 31 parameters of texture feature obtained by CT feature extraction, 19 statistically differed between the two groups. Three independent parameters related to lung adenocarcinoma invasiveness were obtained after dimension reduction, which were angular second moment, contrast, and inverse difference moment. The areas under the curves of the three parameters for diagnosing the invasiveness of early lung adenocarcinoma were 0.715, 0.720, and 0.736, respectively. Conclusion CT texture analysis is helpful in the diagnosis of invasiveness of early lung adenocarcinoma. Key words: Texture analysis; Lung neoplasms; Neoplasm invasiveness; Tomography, X-ray computed