Application of Gray-scale Texture Feature in the Diagnosis of Pulmonary Nodules on CT Imaging

Objective:To explore the value of image texture analysis in the diagnosis of pulmonary nodules.Methods:104 cases of pulmonary nodules were collected,all of which were confirmed by pathology.The level set model was used to extract and segment the lung nodule automatically.The features of energy,contrast,correlation,homogeneity and entropy in pulmonary nodules were calculated by gray-level co-occurrence matrix,which were compared between the benign and malignant lung nodules.Results:The characteristics of energy,contrast,correlation,homogeneity and entropy have significant statistical difference between benign and malignant pulmonary nodules(P＜ 0.05).The means of energy,homogeneity,correlation in benign nodules are greater than that in malignant nodules while the means of contrast and entropy in benign nodules are less than that of malignant nodules.Conclusion:The texture characteristics in pulmonary benign nodules are gray relatively uniform,local texture regular,relatively little contrast in gray,while that in pulmonary malignant nodules are local texture tangle some,gray uneven distribution,marked gray contrast,so the image texture features have important value in the diagnosis of lung nodules.