Preoperative Diagnostic Value Of Low-dose CT Based Radiomic Analysis For Lymph Node Metastasis In Lung Cancer

Objective:To explore the diagnostic value of low-dose CT based radiomics for lymph node?LN?metastasis in lung cancer.Methods:A total of 109 lung cancer patients[LN?+?group with 55 patients:LN?-?group with 54 patients]who underwent surgery or LN biopsy with pathological results from January 2017 to December 2018 were included in this retrospective study.The ¹⁸F-fluoro-deoxyglucose Positron Emission Tomography/Computed Tomography?18F-FDG PET/CT?examination was performed before treatment according to clinical schedule.IntelliSpace Discovery post-processing software was used to sketch the volume of interests?VOIs?on low-dose CT images,and Pyradiomics 2.1 software package was used to extract the radiomic features of each VOI one by one based on the low-dose CT images of PET/CT scan.According to the ratio of 7:3,VOIs were randomly divided into the training group and the verification group.In the training group,the least absolute shrinkage and selection operator?LASSO?method was used for radiomics features selection.The logistic regression analysis method stepwise backwards method used to establish the radiomics model.Finally,in the validation group,the model performance was evaluated by the area under ROC curve?AUC?,calibration curve.Results:There was no significant difference in clinical and pathologicalcharacteristics between the LN metastasis group and non-metastatic group the?P>0.05?.A total of 251 LNs?benign LNs=132;malignant LNs=119?.1472 radiomics features were extracted for each VOI,of which 15radiomics features significantly correlated with LN status selected byLASSO method and logistic regression analysis were used to develop the prediction model.Rad-Score has high predictive efficiency with the AUC[95%Confidence interval?CI?],accuracy,sensitivity,specificity of0.936?0.901-0.970?,0.886,0.904,0.870 in the training group,and 0.875?0.793-0.957?,0.829,0.806,0.850 in the validation group.Conclusion:The radiomics based on low dose chest CT can accurately and noninvasively predict LN metastasis of NSCLC to guide N staging and helped clinicians guide decisions.