A model of malignant risk prediction for solitary pulmonary nodules on 18F-FDG PET/CT:building and estimating

Objective To develop a model of malignant risk predictionof solitary pulmonary nodules ( SPN) with themetabolic characteristics of the lesion. Methods A totalof 362 patients (291 malignant cases and 71 benign cases;194 males,168 females;median age: 61 years)who underwent PET /CT imaging from January 2013 toJuly 2017 were analyzed. The diagnosis of malignantSPN was based on pathological results,and that of benignSPN was based on pathological or follow-up results.Differences of clinical /imaging characteristics in patientswith benign and those with malignant SPN were analyzed.Risk factors were screened by multivariate nonconditionallogistic regression analysis. The self-verificationof the model was done by the receiver operatingcharacteristic (ROC) curve analysis,out-of-group verificationwas performed by k-fold cross-validation.