The predictive value of multiparametric radiomic model for molecular subtypes of adult diffuse gliomas
Objective To explore the predictive value of a multiparametric radiomic model based on MRI images for molecular subtypes of adult diffuse gliomas including isocitrate dehydrogenase(IDH)mutation,IDH wild type,and IDH mutant 1p/19q co-deletion subtypes.Methods The clinical data and preoperative multiparametric MRI data of patients with diffuse gliomas pathologically diagnosed after craniotomy at the Department of Neurosurgery,the First Affiliated Hospital of Zhengzhou University(dataset with 1:465 cases)and the Department of Neurosurgery,Henan Provincial People's Hospital(dataset with 2:137 cases)from January 2011 to September 2021 were collected.The optimal radiomic features were extracted and selected.Then,the radiomic model was constructed using combined algorithms including Boruta and Random Forest.Dataset 1 was randomly divided into a training set and an internal validation set according to the proportion of 7∶3 for machine learning.And dataset 2 was an independent external validation set.The receiver operating characteristic(ROC)curve was drawn,and the area under curve(AUC),sensitivity,specificity values were evaluated to assess the classification performance of radiomic model predicting three molecular subtypes of adult diffuse gliomas.Results Among the dataset 1,there were no significant differences in sex,age,molecular subtypes of gliomas,or WHO grade between the training set(332 cases)and the internal validation set(133 cases,all P＞0.05).And there were no significant differences for above clinical datas between dataset 1(465 cases)and dataset 2(137 cases,all P＞0.05).A total of 8 texture features and 13 intensity features from MRI were significantly correlated with the molecular subtypes of adult diffuse gliomas.The AUC values of IDH mutation,IDH wild type,IDH mutant 1 p/19q co-deleted subtypes of the internal validation set were 0.81,0.86,and 0.85,respectively.The sensitivity values were 89.5％,69.4％,and 75.0％,respectively.And the specificity values were 63.2％,91.4％,and 82.9％,respectively.In the external validation set,the AUC values of IDH mutation,IDH wild type,and IDH mutant 1p/19q co-deleted subtypes were 0.76,0.82,and 0.81,respectively.The sensitivity values were 75.0％,67.3％,and 72.2％,respectively.And the specificity values were 64.6％,78.6％,and 79.8％,respectively.Conclusion The multiparametric radiomic model has high value in prediction of molecular subtypes of adult diffuse gliomas.
万方数据
维普
