Application Of CT Radiomics In Epidermal Growth Factor Receptor Mutation And Pathological Grading Of Non-small Cell Lung Cancer

Part one:Mutation of epidermal growth factor receptor in non-small cell lung cancer based on CT radiomicsObjective: Study of epidermal growth factor receptor based on Computed Tomography(CT)imaging model in Non-small cell lung cancer(NSCLC);EGFR) predictive value in mutation states.Methods: The clinical and imaging data of 125 patients with non-small cell lung cancer diagnosed in our hospital from August 2018 to January 2023 with EGFR gene test results were analyzed by regression.Among them,32 patients with EGFR mutation,including 11 males and 21 females,were(60.81±9.38)years old.There were 93 cases of EGFR wild-type,including 42 males and 51 females,with an average age of(62.14±8.86)years.Patients were randomly divided into test set and training set in a ratio of 1:4.Clinical information,including sex,age,smoking status,and Carcino-embryonic antigen(CEA)level,were recorded.The CT features of each patient,such as lobulation,burr sign and ground glass density,were analyzed and described in detail.High standard manual segmentation of the captured CT images can be achieved with the software of Hui Medical Hui Image.Moreover,through the use of Least absolute shrinkage and selection operator,Lasso’s method was used to select the best feature subset and construct the radiomics model.area under the curve,area under the curve,Receiver Operating Characteristic(ROC)was drawn.AUC)to evaluate the predictive power of radiomics features in EGFR mutations in non-small cell lung cancer.Results: A total of 1409 features were extracted from the EGFR mutation prediction model of non-small cell lung cancer.After screening,five optimal feature subsets were obtained to construct the model.The established radiomics model could effectively predict the EGFR mutation status.The training set AUC was 0.80(95%CI:0.70-0.91),and the specificity and sensitivity were 0.72 and 0.72,respectively.The AUC of the test set was 0.75(95%CI: 0.53-0.97),and the specificity and sensitivity were 0.71 and 0.74,respectively.Conclusion: CT imaging model has good predictive value in epidermal growth factor receptor mutation state of non-small cell lung cancer.Part two:Application of CT radiomics in pathological grading of non-small cell lung cancerObjective: To investigate the predictive value of CT imaging model in the pathological grading of non-small cell lung cancer.Materials and METHODS: Data of 67 patients with pathologically confirmed nonsmall cell lung cancer with definite pathological grade from June 2018 to January 2021 were retrospectively analyzed.Among them,there were 7 cases of grade Ⅰ,39 cases of grade Ⅱ,and 21 cases of grade Ⅲ.According to the degree of differentiation,atypia,and number of mitotic images of small cell lung cancer,grade Ⅰ and Ⅱ were classified as low-grade group(46 cases in total),and grade Ⅲ was classified as highgrade group(21 cases in total).The patients were randomly divided into a training group(56 cases)and a test group(11 cases).The 256-row Revolution CT scanner produced by General Electric(GE)was used for lung scanning.Darwin intelligent research platform software was used to select the maximum diameter of the lesions and conduct Region of interest.A total of 1878 image omics features were obtained by manual delineation of ROI and extraction of all focal features outlined.The features were screened by maximum and minimum normalization and variance analysis,and 20 optimal feature subsets were finally obtained.The prediction model was established by support vector machine(SVM).The predictive value of the imaging features in the pathological grading of non-small cell lung cancer was evaluated by drawing the receiver operating characteristic curve and calculating the area under ROC curve.Results: In the training group,the area under ROC curve was 0.851(95%CI: 0.68-1.00),and the specificity and sensitivity were 92.31% and 70.59%,respectively.The area value under ROC curve in the test group was 0.83(95%CI: 0.28-1.00),and the specificity and sensitivity were 85.20% and 71.35%,respectively.Conclusion: CT radiomics model has certain reference value for evaluating different pathological grades of non-small cell lung cancer,and has certain guiding significance for diagnosis and qualitative of lung cancer.