The Value Of MRI Radiomics In Differential Diagnosis Of High-Grade Gliomas And Solitary Brain Metastases

PART 1The Value of MRI Radiomics in Differential Diagnosis of High-Grade Gliomas and Solitary Brain MetastasesObjective:To explore the value of MRI radiomics in differential diagnosis of High-Grade Gliomas(HGG)and Solitary Brain Metastases(SBM).Methods:The MRI imaging data of 52 patients with high-grade glioma and 72 patients with brain metastatic tumor who were treated in our hospital from January2012 to March 2022 in Yijishan Hospital of Wannan Medical College were retrospectively analyzed.Patients were divided into the training group(n=86)and the test group(n=38)in a ratio use a completely randomized method.Clinical date were collectd.Tumor regions on all MRI images were registered to and contoured in T2-weighted fluid-attenuated inversion recovery(FLAIR)images.Subtraction of T1WI(T1 weighted imaging)and T1/C(T1WI enhancement sequence)images in the same slice of lesion to constract the subtraction images.The MRI images of all patients were imported into ITK-SNAP software to manually sketch tumor regions of interest and extract radiomics features..The pearson correlation coefficients and kruskal-wallis test were used to reduce the dimension and select the feature,then the Single sequence MRI(T1WI、T2WI、T2-FLAIR、DWI、ADC、T1/C、subtraction images)and multi-sequence MRI(plain scan、plain scan+T1/C、plain scan+subtraction、plain scan+T1/C+subtraction)radiomics model was established.Integrated the most effective radiomics model and the clinical features to construct the synthetic diagnostic model.The area under the receiver operating characteristic curve(AUC)was used to evaluate the diagnostic efficacy of the model and the clinical usefulness was assessed by decision curve analysis(DCA).Results: The diagnostic performance of T1/C subtraction images radiomics model(training set AUC vs test set AUC : 0.929vs0.952)is better than the others(T1WI radiomics model : 0.918vs0.912;T2WI radiomics model: 0.906vs0.886;T2-FLAIR radiomics model: 0.934vs0.926;DWI radiomics model: 0.928vs0.923;ADC radiomics model:0.934vs0.906;T1/C radiomics model: 0.927vs0.915)in the Single sequence MRI radiomics model;The diagnostic performance of the multi-sequence MRI images radiomics model(training set AUC vs test set AUC : 0.933vs0.946)is better than the others(non-enhanced MRI images radiomics model training set AUC vs test set AUC : 0.924vs0.918,non-enhanced MRI images and T1/C model training set AUC vs test set AUC : 0.931vs0.906)In the combined MRI radiomics model.Compared with the clinical model(training set AUC: 0.744;test set AUC: 0.744)and the multi-sequence MRI images radiomics model,the synthetic diagnostic model had better diagnosis effects in both the training set(AUC: 0.942)and the test set(AUC:0.992).Conclusions: The MRI radiomics model could noninvasively and efficiently Differential Diagnosis of High-Grade Gliomas and Solitary Brain Metastases.PART 2The Value of Intra-/Peri-tumoral MRI radiomics in Differential Diagnosis of High-Grade Gliomas and Solitary Brain MetastasesObjective:To explore the value of intra-/peri-tumoral MRI radiomics in differential diagnosis of high-grade gliomas and solitary brain metastases.Methods:The MRI imaging data of 52 patients with high-grade glioma and 72 patients with brain metastatic tumor who were treated in our hospital from January2012 to March 2022 in Yijishan Hospital of Wannan Medical College were retrospectively analyzed.Patients were divided into the training group(n=86)and the test group(n=38)in a ratio use a completely randomized method.Clinical date were collectd.Tumor regions on all MRI images were registered to and contoured in T2-weighted fluid-attenuated inversion recovery(FLAIR)images.The MRI images of all patients were imported into ITKSNAP software to manually sketch tumor regions of interest.The RIAS were used to expand 3mm along the edge of the tumor.Radiomic features were respectively extracted from the intratumoral ROI and peritumoral ROI.The radiomic features of intratumoral ROI peritumoral ROI were combined to generate fusion radiomic feature.The area under the receiver operating characteristic curve(AUC)was used to evaluate the diagnostic efficacy of the model,And the clinical usefulness was assessed by decision curve analysis(DCA).Results: The predictive efficient of peritumoral radiomic model AUC(training set vs test set :0.937vs0.892、0.966vs0.928、0.946vs0.920)was higher than intratumoral radiomic model(training set vs test set :0.919vs0.898、0.916vs0.915、0.927vs0.884)and fusion radiomic model(training set vs test set :0.937vs0.892、0.946vs0.920、0.936vs0.878)which based on T1WI、T2WI、T2-FLAIR。The predictive efficient of peritumoral radiomic model based on T2 WI is the highest.Conclusions: The T2 WI peritumoral radiomic model provides reliable and quantified objective basis for the differential diagnosis HGG and SBM.The non-enhanced MRI images peritumoral radiomic model has a higher predictive performance and a Potential application value.