Objective To explore the value of MRI radiomics based on the frontal and temporal lobes combined with multiple machine learning algorithms in the diagnosis of mild cognitive impairment(MCI).Methods Patients who underwent cranial MR examination were retrospectively selected.According to the inclusion and exclusion criteria,a total of 173 subjects were finally included and randomly divided into training set and test set in a ratio of 7∶3.After delineating the regions of interest(ROI)of the frontal and temporal lobes on T2-fluid attenuated inversion recovery(FLAIR)images,radiomics features were extracted based on the Pyradiomics data package.Features were screened through inter-and intraclass correlation coefficient(ICC),independent samples t-test,and the LightGBM algorithm.Diagnostic models were constructed using support vector machine(SVM),random forest(RF),decision tree(DT),K-nearest neighbor(KNN),gradient boosting decision tree(GBDT),and extreme gradient boosting(XGBoost)combined with 10-fold cross-validation respectively.The training set was further divided into 9 training data sets and 1 validation data set through 10-fold cross-validation,and the hyperparameters were optimized through iterative cycles.The diagnostic efficacy of the model was evaluated by receiver operating characteristic(ROC)curve and area under the curve(AUC),and the DeLong test was applied to compare the differences between different models.Results The AUC of the radiomics models constructed by SVM,DT,RF,KNN,GBDT,XGBoost in the training set were 0.951,0.992,0.998,0.957,1.000,and 1.000 respectively,in the validation set were 0.890,0.843,0.934,0.878,0.930,and 0.945 respectively,and in the test set were 0.902,0.711,0.899,0.849,0.889,and 0.882 respectively.Conclusion MRI radiomics based on the frontal and temporal lobes combined with multiple machine learning algorithms can diagnose MCI,and the model constructed based on SVM shows the highest diagnostic value.

Objective To explore the value of MRI radiomics based on the frontal and temporal lobes combined with multiple machine learning algorithms in the diagnosis of mild cognitive impairment(MCI).Methods Patients who underwent cranial MR examination were retrospectively selected.According to the inclusion and exclusion criteria,a total of 173 subjects were finally included and randomly divided into training set and test set in a ratio of 7∶3.After delineating the regions of interest(ROI)of the frontal and temporal lobes on T2-fluid attenuated inversion recovery(FLAIR)images,radiomics features were extracted based on the Pyradiomics data package.Features were screened through inter-and intraclass correlation coefficient(ICC),independent samples t-test,and the LightGBM algorithm.Diagnostic models were constructed using support vector machine(SVM),random forest(RF),decision tree(DT),K-nearest neighbor(KNN),gradient boosting decision tree(GBDT),and extreme gradient boosting(XGBoost)combined with 10-fold cross-validation respectively.The training set was further divided into 9 training data sets and 1 validation data set through 10-fold cross-validation,and the hyperparameters were optimized through iterative cycles.The diagnostic efficacy of the model was evaluated by receiver operating characteristic(ROC)curve and area under the curve(AUC),and the DeLong test was applied to compare the differences between different models.Results The AUC of the radiomics models constructed by SVM,DT,RF,KNN,GBDT,XGBoost in the training set were 0.951,0.992,0.998,0.957,1.000,and 1.000 respectively,in the validation set were 0.890,0.843,0.934,0.878,0.930,and 0.945 respectively,and in the test set were 0.902,0.711,0.899,0.849,0.889,and 0.882 respectively.Conclusion MRI radiomics based on the frontal and temporal lobes combined with multiple machine learning algorithms can diagnose MCI,and the model constructed based on SVM shows the highest diagnostic value.