Radiomics analysis of 18F-FDG PET/CT for visceral pleural invasion in non-small cell lung cancer with pleural attachment

Abstract

Abstract Objective The aim of this study was to establish and validate a preoperative model that integrates clinical factors and radiomic features from 18F-FDG PET/CT for the prediction of visceral pleural invasion (VPI) in non-small-cell lung cancer (NSCLC) with pleural attachment. Methods A total of 814 NSCLC patients with radiological pleural attachment were included in this retrospective study. VPI was confirmed in 350 cases, and non-VPI was confirmed in 464 cases through histopathological examination. The patients were randomly divided into a training set and a test set. Clinical data and 101 radiomic features (51 PET features and 50 CT features) were collected. The optimal predictors from these radiomic features were selected using the Mann-Whitney U test and the least absolute shrinkage and selection operator (LASSO) algorithm, resulting in the creation of the radiomics score (Rad-score) for the PET/CT radiomics model. Meaningful clinical factors and Rad-scores were incorporated into a combined PET/CT radiomics-clinical model through multivariate logistic regression analysis. The predictive performance and clinical utility of the models were assessed using receiver operating characteristic (ROC) analysis and decision curve analysis (DCA). Finally, a nomogram was developed based on the best-performing model. Results The combined PET/CT radiomics-clinical model to predict VPI status achieved the areas under the ROC curve (AUCs) of 0.840, 0.890, and 0.884, in the training set (n = 569), test set (n = 245), and patients with a maximum tumor diameter (Dmax) ≤ 3 cm (n = 437), respectively, which were significantly higher than 0.763, 0.747, and 0.813 of the clinical model, and 0.723, 0.763, and 0.719 of the PET/CT radiomics model. The DCA showed that the combined model had the highest standardized net benefit among the models in predicting VPI. Subsequently, a nomogram based on the combined model was developed with well-fitted calibration curves. Conclusions The combined PET/CT radiomics-clinical model offers an advantage in the prediction of VPI in NSCLC with pleural attachment.