The Value Of Multi-Parametric MRI-Based Radiomics Model In Predicting The IDH1 Mutation And 1p/19q Codeletion In Glioma

Objective To explore the value of multi-parametric MRI-based radiomics model for predicting IDH1 mutation and 1p/19 q codeletion in glioma.Methods A total of 186 patients with diffuse gliomas confirmed by pathology from January 2016 to July 2022 in our hospital were retrospectively included.There were 88 patients with IDH1 mutation and 98 patients with IDH1 wild type,37 patients with 1p/19 q codeletion and 51 patients with 1p/19 q non-codeletion.Preoperative MRI images of patients were collected,including T2-weighted imaging(T2WI),T2-weighted fluid-attenuated inversion recovery(T2-FLAIR)and contrast-enhanced T1-weighted imaging(CE-T1WI).All patients were randomly divided into training and testing sets at a 7:3 ratio.ITK-SNAP software was used to delineate the region of interest(ROI)and the radiomics features were extracted by the "Pyradiomics" package in Python 3.7 software.The process of feature screening and dimension reduction were as follows: firstly,the independent Student’s t-test or the Mann-Whitney U test was used to initially screen out the features with statistically significant differences(P<0.05),then the Pearson correlation analysis and the least absolute shrinkage and selection operator(LASSO)were used to select the optimal feature which were highly correlated with IDH1 mutation and 1p/19 q codeletion to construct radiomics labels and calculate the radiomic score(Rad-score).Finally,we used the logistic regression algorithm to construct three models from the T2 WI,T2-FLAIR,and CE-T1 WI sequences respectively as well as a combined model using them all.The performance was evaluated using area under the receiver operating characteristic(ROC)curves(AUC)and calibration curves,the radiomics nomogram would also be constructed.Results1.Comparison of clinical and MRI imaging features: There were statistically differences in patient’s age(P=0.010),enhancement degree(P<0.001),oedema degree(P<0.001),enhancement style(P<0.001),Ki-67 expression level(P<0.001)and WHO-grade(P<0.001)between IDH-mutated and IDH-wild groups.There were statistically differences in tumor boundary(P=0.001)and the frontal lobe(P=0.024)between 1p/19 q codeletion and non-codeletion groups.2.When predicting IDH1 mutation,the CE-T1 WI model had the best prediction performance among single-sequence radiomics models.Its AUC values in the training and testing groups were 0.828(95%CI: 0.751～0.896)and 0.776(95%CI: 0.649～0.893).The diagnostic efficiency of the combined model was better than all the single-sequence radiomics models,its AUC values in the training and testing groups were 0.840(95%CI:0.761～0.907)and 0.830(95%CI:0.716～0.921).3.When predicting 1p/19 q codeletion,the T2-FLAIR model had the best prediction performance among single-sequence radiomics models.Its AUC values in the training and testing groups were 0.803(95%CI:0.683～0.901)and 0.794(95%CI:0.583～0.939).The diagnostic efficiency of the combined model was better than all the single-sequence radiomics models,its AUC values in the training and testing groups were 0.824(95%CI:0.691～0.923)and 0.830(95%CI:0.656～0.965).Conclusion Multi-parametric MRI-based radiomics model could non-invasively predict IDH1 mutation and 1p/19 q codeletion in glioma,however,the performance of different sequences is different.It was shown that among various single-sequence radiomics models,the CE-T1 WI radiomics model should be considered as an optimal model in predicting IDH1 mutation,the T2-FLAIR radiomics model should be considered as an optimal model in predicting 1p/19 q codeletion while the combined model based on three sequences could further improve the predicting performance.