Comparison of T1W-enhanced radiomics model and multi-parameter MRI model:predicting isocitrate dehydrogenase mutation status in glioma
DU Ningfang
ZHANG Xiao
LU Yawen
SUN Lianxi
YANG Danping
LI Shihong
Abstract：Objective To construct and validate a T1W-enhanced radiomics model,and compare its preoperative predictive efficacy for the isocitrate dehydrogenase(IDH)mutation status in glioma with that of the multi-parameter MRI and the combined model.Methods Retrospective analysis was conducted on the clinical data and multiparametric MRI features of 127 patients with pathologically confirmed glioma in Huadong Hospital Affiliated to Fudan University from 2020 to 2023.All patients were randomly divided into a training group and a validation group at a ratio of 8:2.Regions of interest(ROI)were delineated on preoperative T1W-enhanced images,and radiomic features were extracted and screened using Pyradiomics software,ultimately obtaining 5 optimized radiomic features.Combined with clinical data and multiparametric MRI features,3 logistic models were constructed,namely the T1W-enhanced radiomics model,the multi-parameter MRI model,and the combined model.The performance of the models was evaluated using ROC curves and the area under the curve(AUC).Results Among the three predictive models for the IDH mutation status in glioma,the T1W-enhanced radiomics model exhibited the best predictive efficacy.The AUC values of the training group and validation group were 0.860(95%CI:0.783-0.937)and 0.955(95%CI:0.880-1.000),respectively.The Delong test demonstrated that the T1W-enhanced radiomics model had superior predictive performance compared to the multi-parameter MRI model(P=0.011).However,compared with the T1W-enhanced radiomics model,the combined model failed to further improve the predictive efficacy for the IDH mutation status in glioma(P=0.067).Conclusion The preoperative T1W-enhanced radiomics model can be used to predict the IDH genotype of glioma,with higher predictive efficacy than the traditional multi-parameter MRI model.
Keywords：gliomaisocitrate dehydrogenasemagnetic resonance imagingradiomicsmachine learning
Publication Date：2026-02-20
Online Publishing Date：2026-03-25(First online date of this platform, not the publication date of the document)
Pages：9( 211-219 )