Objective To explore the predictive value of CT-based radiomics nomogram for Ki-67 expression level in non-small cell lung cancer(NSCLC).Methods The clinical and chest CT imaging data of 144 patients with NSCLC confirmed by pathology and tested for Ki-67 expression level were retrospectively analyzed.The patients were randomly divided into training group(100 cases)and validation group(44 cases).According to the expression level of Ki-67 in the pathological report,NSCLC patients were divided into low expression group(Ki-67＜14％)and high expression group(Ki-67＞14％).In the training group,the clinical characteristics and CT signs of patients with low and high Ki-67 expression were analyzed,and univariate and multivariate Logistic regression analysis were used to screen out independent predictors and establish a clinical model.Radiomics features were extracted from chest plain CT lung window images,and the maximum absolute value normalization,optimal feature selection(percentage),and model selection and selection operator(LASSO)algorithm were used to reduce the dimension of the data.The radiomics score(Rad-score)was calculated and the radiomics model was established.The independent predictors and radiomics scores were analyzed by Logistic regression to obtain a combined nomogram model.ROC curve and area under the curve(AUC)were used to evaluate the predictive efficacy of the three models.Results ROC curve analysis of the data in the training group and validation group showed that the combined nomogram model had the largest AUC of 0.873(95％CI:0.791-0.931)and 0.851(95％CI:0.712-0.940),respectively.Compared with the clinical model and radiomics model,the combined nomogram model had the largest AUC of 0.873(95％CI:0.791-0.931)and 0.851(95％CI:0.712-0.940),respectively.It has a better predictive value for Ki-67 expression level in NSCLC.Hosmer-Lemeshow test showed that the combined model of training group and validation group was consistent with the actual outcomes(P＞0.05).Conclusion The CT-based radiomics nomogram provides a method for preoperative non-invasive prediction of Ki-67 proliferation index in non-small cell lung cancer,which makes the evaluation of tumor differentiation more optional,and can provide supplementary information for clinicians and select appropriate treatment plans.
维普
万方数据