Predictive Value Of Radiomics For Invasiveness Of Ground-glass Pulmonary Nodules And Distant Metastasis Of Solid Lung Adenocarcinoma

Part Ⅰ: Predictive value of radiomics for the invasiveness of ground-glass pulmonary nodulesObjectives: To develop,validate and compare a clinical model,a volume model,and a radiomics model based on unenhanced chest CT to identify precursor glandular lesions and adenocarcinomas in ground-glass nodules(GGNs).Methods: This retrospective study included 199 patients with 247 GGNs from January 2018 to September 2022.They are divided into training set and test set according to the order of diagnosis time with a ratio of 7:3.The region of interest(ROI)of the GGNs was automatically segmented,and 3 volume features and 96 radiomics features were extracted using the intelligent auxiliary diagnostic system of Shukun’s chest CT.Logistic regression models were trained separately with clinical,volumetric,and radiomics features.The receiver operating characteristic curve(ROC),calibration curve,and decision curve analysis(DCA)were used to evaluate the model.The Delong test was used to compare the area under curve(AUC)between models.The Akaike information criterion(AIC)comprehensively considered the parsimony and goodness of fit of the models.In addition,this study also analyzed the performance of each model in subgroups with different image slice thicknesses.Results: The AUCs of the clinical model,volume model,and radiomics model in the test set were 0.905,0.906,and 0.893,respectively.The Delong test showed that there was no statistically significant difference in AUC values among the three models(p > 0.05).The ROC curve,calibration curve,DCA and AIC values of the three models were slightly different,but the overall difference was not significant.The results of subgroup analysis showed that the AUCs of the clinical model,volumetric model,and radiomics model in the subgroup with thickness of 1.00 mm were 0.877,0.876,and 0.877,respectively,and the AUCs in the subgroup with thickness of 1.25 mm were 0.823,0.858,and 0.848,respectively,the De Long test showed no statistical difference.Conclusion: The models based on clinical,volumetric,and radiomics features have good performance and clinical value in predicting invasiveness of GGNs,and their performance is similar.The predictive performance of the clinical,volumetric,and radiomics models decreased in the subgroup with a slice thickness of 1.25 mm,but the difference was not significant compared with a slice thickness of 1.00 mm.Part Ⅱ: Predictive value of radiomics in distant metastasis of solid lung adenocarcinomaObjectives: To develop a comprehensive model based on three-dimensional(3D)radiomic features,two-dimensional(2D)radiomic features,and clinical features of the chest CT enhanced scans to predict distant metastases in patients with solid lung adenocarcinomas before treatment.Methods: This retrospective study included 253 eligible patients with solid adenocarcinoma of the lung diagnosed at our hospital between August 2018 and August 2021.They are divided into training set and test set according to the order of diagnosis time with a ratio of 7:3.3D and 2D regions of interest were segmented from computed tomography-enhanced thin-slice images of the venous phase,and 851 radiomic features were extracted in each region.The Least Absolute Shrinkage and Selection Operator(LASSO)was used to select radiomic features and calculate radiomic scores,and logistic regression was used to develop the model.Development of a 3D radiomics model,a 2D radiomics model,a combined 3D and 2D radiomics model,a clinical model,and a comprehensive model for the prediction of distant metastases in patients with solid lung adenocarcinomas.Nomograms were drawn to illustrate comprehensive model,and ROC curve,calibration curve,and DCA were used for model evaluation.Results: The AUC of 3D radiomics model,2D radiomics model,combined 3D and 2D radiomics model,clinical model,and comprehensive model in the test set was0.711,0.769,0.775,0.829 and 0.892,respectively.The Delong test showed that AUC values were statistically different between comprehensive model and 3D radiomics model(p=0.001),and there was no statistical difference in AUC between the other models.Based on a comprehensive review of DCA,ROC curve,and AIC,Comprehensive model is demonstrated to have better clinical utility,goodness of fit,and parsimony.Conclusion: A comprehensive model based on 3D radiomics features,2D radiomics features,and clinical features has the potential to predict distant metastasis in patients with solid lung adenocarcinomas.The predictive performance of 2D radiomics models was comparable to that of 3D radiomics models in predicting distant metastasis of solid adenocarcinoma of the lung.