Logistic regression analysis of CT characteristics of solitary pulmonary nodules

Objective: To determine the CT characteristics of benign and malignant solitary pulmonary nodules (SPN) using logistic regression analysis. Methods: Of 186 histologically confirmed SPN including 125 primary lung cancer and 61 benign nodules, 72 malignant and 28 benign nodules were included in the study using completely randomized method. CT features of benign and malignant SPN including location, size, border, internal structure, relationship with surrounding blood vessels and pleura were compared using χ² test and logistic regression analysis. Results: There was no significant difference in the lesion location, size, air bronchogram, air alveologram, or hypervascularity between the benign and malignant SPN by χ² test. Using logistic regression analysis, spiculation and pleural adhesion allowed diagnosis of malignant SPN with odds ratios (95% confidence intervals) of 38.529 (6.677-222.336) and 11.963 (1.904-75.183), respectively. The diagnostic accuracy (95.0%), sensitivity (95.8%) and specificity (92.9%) of logistic regression analysis were high. Conclusions: Spiculation and pleural adhesion are CT characteristics of malignant SPN using logistic regression analysis. (authors)