The Study Of The Value And Optimization Of Multiparametric Functional MRI And Radiomics In The Prediction Of IDH1 Mutation Status 1p/19q Codeletion Status And Proliferation Of Gliomas

Chapter I The Study of the Value and Parameter Optimization of Multiparametric Functional MRI in the Prediction of IDH1 Mutation Status,1p/19q Codeletion Status,Proliferation and Grading of GliomasPart I The value of three-dimensional arterial spin labeling in the prediction of IDH1 mutation status,lp/19q codeletion status,proliferation and grading of gliomasObjective:To explore the value of three-dimensional arterial spin labeling(3D-ASL)in the prediction of IDH1 mutation status,1p/19q codeletion status,proliferation and grading of gliomas.Methods:All the patients diagnosed with suspicious supratentorial adult-type diffuse gliomas underwent conventional MRI(including T2WI,T2 FLAIR and CE T1WI)and 3D-ASL before operation,and 92 patients were confirmed by histopathology.The ASL parameter-cerebral blood flow(CBF)was compared between IDH1 mutation and IDH1 widetype groups,1p/19q codeletion and 1p/19q non-codeletion groups,low Ki-67 and high Ki-67 groups,astrocytoma,IDH-mutant WHO 2 and WHO 3-4 groups,and oligodendroglioma,IDH-mutant,and 1p/19q-codeleted WHO 2 and WHO 3 groups using Mann-Whitney U test,and P<0.05 was regarded as statistically significant.Receiver operating characteristic(ROC)curve was used to assess the diagnostic efficacy of CBF.Results:The CBF of IDH1 wildtype,1p/19q non-codeletion and high Ki-67 groups was higher than that of IDH1 mutation,1p/19q co-deletion and low Ki-67 groups,respectively(P<0.05).The areas under the curve(AUCs)for identifying IDH1 mutation and IDH1 widetype groups,1p/19q codeletion and 1p/19q non-codeletion groups,and low Ki-67 and high Ki-67 groups by CBF value were 0.65,0.73 and 0.634,respectively,with accuracy of 69.6%,76.8%and 65.2%,sensitivity of 57.8%,78.3%and 81.6%,specificity of 80.9%,86.7%and 41.9%,and cut-off value of 68.189,66.893 and 62.646.There was no significant correlation between CBF and grading of gliomas.Conclusion:3D-ASL can be used for the prediction of IDH1 mutation status,lp/19q co-deletion status and proliferation of gliomas,but not grading.Part II The value of multi-b value diffusion-weighted imaging in the prediction of IDH1 mutation status,1p/19q codeletion status,proliferation and grading of gliomasObjective:To explore the value of multi-b value diffusion-weighted imaging(DWI)in the prediction of IDH1 mutation status,1p/19q codeletion status,proliferation and grading of gliomas,and select the best parameters.Methods:All the patients diagnosed with suspicious supratentorial adult-type diffuse gliomas underwent conventional MRI and multi-b value DWI before operation,and 92 patients were confirmed by histopathology.By using bi-exponential model slow ADC,fast ADC and fraction of fast ADC were calculated,and using stretched-exponential model DDC and Alpha were calculated.These parameters were compared between IDH1 mutation and IDH1 widetype groups,1p/19q codeletion and 1p/19q non-codeletion groups,low Ki-67 and high Ki-67 groups,astrocytoma,IDH-mutant WHO 2 and WHO 3-4 groups,and oligodendroglioma,IDH-mutant,and 1p/19q-codeleted WHO 2 and WHO 3 groups using Mann-Whitney U test respectively.P<0.05 was regarded as statistically significant.ROC curve was used to assess the diagnostic efficacy of statistically significant data,and multivariate logistic regression analysis was used to select the best function parameters.Results:The combination of all the multi-b value DWI parameters was different significantly between IDH1 mutation and IDH1 widetype groups,1p/19q codeletion and 1p/19q non-codeletion groups,and low Ki-67 and high Ki-67 groups(P<0.05),with AUCs of 0.702,0.766 and 0.847,accuracy of 66.3%,68.5%and 77.2%,sensitivity of 84.4%,78.3%and 83.7%,specificity of 48.9%,65.2%and 69.8%,respectively.There was no significant correlation between multi-b value DWI and grading of gliomas.The slow ADC,DDC and Alpha values were significantly higher in IDH1 mutation group than in IDH1 widetype group,whereas the fraction of fast ADC values were significantly lower in IDH1 mutation(P<0.05),with AUCs of 0.702,0.68,0.703 and 0.667,respectively.Slow ADC was the best function parameter in differentiating between IDH1 mutation and IDH1 widetype gliomas.There was no significant correlation between fast ADC and IDH1 mutation status of gliomas.The slow ADC,fraction of fast ADC,DDC and Alpha values were significantly higher in lp/19q codeletion group than in lp/19q non-codeletion group,whereas the fast ADC values were significantly lower in 1p/19q codeletion group(P<0.05),with AUCs of 0.717,0.702,0.69,0.721 and 0.654,respectively.Slow ADC,fraction of fast ADC and DDC were the best function parameters in differentiating between 1p/19q codeletion and 1p/19q non-codeletion gliomas.The slow ADC,DDC and Alpha values were significantly higher in low Ki-67 group than in high Ki-67 group,whereas the fast ADC and fraction of fast ADC values were significantly lower in low Ki-67 group(P<0.05),with AUCs of 0.818,0.775,0.746,0.621 and 0.742,respectively.Slow ADC,fraction of fast ADC and DDC were the best function parameters in differentiating between low Ki-67 and high Ki-67 gliomas.Conclusion:Multi-b value DWI can be used for the prediction of IDH1 mutation status,1p/19q codeletion status and proliferation of gliomas,but not grading.And bi-exponential model was better than stretched-exponential model,in which slow ADC was the best function parameter.Part III The value of diffusion-kurtosis imaging in the prediction of IDH1 mutation status,lp/19q codeletion status,proliferation and grading of gliomasObjective:To explore the value of diffusion-kurtosis imaging(DKI)in the prediction of IDH1 mutation status,1p/19q codeletion status,proliferation and grading of gliomas,and select the best parameters.Methods:All the patients diagnosed with suspicious supratentorial adult-type diffuse gliomas underwent conventional MRI and DKI before operation,and 92 patients were confirmed by histopathology.FA,MD,Da,Dr,FAK,MK,Ka and Kr were calculated.These parameters were compared between IDH1 mutation and IDH1 widetype groups,1p/19q codeletion and 1p/19q non-codeletion groups,low Ki-67 and high Ki-67 groups,astrocytoma,IDH-mutant WHO 2 and WHO 3-4 groups,and oligodendroglioma,IDH-mutant,and 1p/19q-codeleted WHO 2 and WHO 3 groups using Mann-Whitney U test respectively.P<0.05 was regarded as statistically significant.ROC curve was used to assess the diagnostic efficacy of statistically significant data,and multivariate logistic regression analysis was used to select the best function parameters.Results:The combination of all the DKI parameters was different significantly between IDH1 mutation and IDH1 widetype groups,1p/19q codeletion and 1p/19q non-codeletion groups,low Ki-67 and high Ki-67 groups and astrocytoma,IDH-mutant WHO 2 and WHO 3-4 groups(P<0.05),with AUCs of 0.658,0.679,0.799 and 0.718,accuracy of 64.1%,63%,72.8%and 71.1%,sensitivity of 73.3%,78.3%,71.4%and 73.9%,specificity of 55.3%,58%,74.4%and 59.1%,respectively.There was no significant correlation between DKI and grading of oligodendroglioma,IDH-mutant,and 1p/19q-codeleted.The MD and Dr values were significantly higher in IDH1 mutation group than in IDH1 widetype group,whereas MK,Ka and Kr values were significantly lower in IDH1 mutation(P<0.05),with AUCs of 0.63,0.64,0.656,0.658 and 0.637,respectively.MK and Ka were the best function parameters in differentiating between IDH1 mutation and IDH1 widetype groups.There was no significant correlation between FA and IDH1 mutation status,Da and IDH1 mutation status,and FAK and IDH1 mutation status of gliomas.The MD and Dr values were significantly higher in 1p/19q codeletion group than in 1p/19q non-codeletion group,whereas FAK,MK,Ka and Kr values were significantly lower in lp/19q codeletion group(P<0.05),with AUCs of 0.648,0.665,0.678,0.684,0.668 and 0.679,respectively.MK and Kr were the best function parameters in differentiating between 1p/19q codeletion and 1p/19q non-codeletion groups.There was no significant correlation between FA and 1p/19q codeletion status,and Da and 1p/19q codeletion status of gliomas.The MD,Da and Dr values were significantly higher in low Ki-67 group than in high Ki-67 group,whereas the FA,FAK,MK,Ka and Kr values were significantly lower in low Ki-67 group(P<0.05),with AUCs of 0.752,0.724,0.758,0.707,0.657,0.813,0.806 and 0.799,respectively.MK and Ka was the best function parameters in differentiating between low Ki-67 and high Ki-67 groups.The Ka value was significantly higher in astrocytoma,IDH-mutant WHO 3-4 group than in WHO 2 group(P<0.05),with AUCs of 0.718.Ka was the best function parameter.Conclusion:DKI can be used for the prediction of IDH1 mutation status,1p/19q codeletion status,proliferation of gliomas and grading of astrocytoma,IDH-mutant.Ka was the best function parameter for the prediction of IDH1 mutation status of gliomas and grading of astrocytoma,IDH-mutant,and MK was the best for 1p/19q codeletion status and proliferation of gliomas.Part Ⅳ The value of amide proton transfer imaging in the prediction of IDH1 mutation status,1p/19q codeletion status,proliferation and grading of gliomasObjective:To explore the value of amide proton transfer(APT)imaging in the prediction of IDH1 mutation status,1p/19q codeletion status,proliferation and grading of gliomas.Methods:All the patients diagnosed with suspicious supratentorial adult-type diffuse gliomas underwent APT imaging before operation,and 92 patients were confirmed by histopathology.The APT values were compared between IDH1 mutation and IDH1 widetype groups,1p/19q codeletion and 1p/19q non-codeletion groups,low Ki-67 and high Ki-67 groups,astrocytoma,IDH-mutant WHO 2 and WHO 3-4 groups,and oligodendroglioma,IDH-mutant,and 1p/19q-codeleted WHO 2 and WHO 3 groups using Mann-Whitney U test,and P<0.05 was regarded as statistically significant.ROC curve was used to assess the diagnostic efficacy of APT values.Results:The APT values of IDH1 wild-type group and high Ki-67 group were higher than that of IDH1 mutation group and low Ki-67 group,respectively(P<0.05).The AUCs for identifying IDH1 mutation and IDH1 widetype groups and low Ki-67 and high Ki-67 groups by APT value were 0.658 and 0.69,with accuracy of 65.2%and 66.3%,sensitivity of 57.8%and 59.2%,specificity of 72.3%and 69.8%,and cut-off value of 2.465 and 2.895,respectively.There was no significant correlation between APT values and 1p/19q codeletion status,and APT values and grading of gliomas.Conclusion:APT can be used for the prediction of IDH1 mutation status and proliferation of gliomas,but not 1p/19q codeletion status and grading.Part Ⅴ The value of susceptibility-weighted imaging in the prediction of IDH1 mutation status,1p/19q codeletion status,proliferation and grading of gliomasObjective:To explore the value of susceptibility-weighted imaging(SWI)in the prediction of IDH1 mutation status,lp/19q codeletion status,proliferation and grading of gliomas.Methods:All the patients diagnosed with suspicious supratentorial adult-type diffuse gliomas underwent SWI before operation,and 92 patients were confirmed by histopathology.Intratumoral susceptibility signal intensity(ITSS)was compared between IDH1 mutation and IDH1 widetype groups,1p/19q codeletion and 1p/19q non-codeletion groups,low Ki-67 and high Ki-67 groups,astrocytoma,IDH-mutant WHO 2 and WHO 3-4 groups,and oligodendroglioma,IDH-mutant,and lp/19q-codeleted WHO 2 and WHO 3 groups using Mann-Whitney U test,and P<0.05 was regarded as statistically significant.ROC curve was used to assess the diagnostic efficacy of ITSS.Results:The ITSS of IDH1 wild-type group,1p/19q non-codeletion group,high Ki-67 group and oligodendroglioma,IDH-mutant,and lp/19q-codeleted WHO 3 group were higher than that of IDH1 mutation group,1p/19q codeletion group,low Ki-67 group and WHO 2 group,respectively(P<0.05).The AUCs for identifying IDH1 mutation and IDH1 widetype groups,lp/19q codeletion and lp/19q non-codeletion groups,low Ki-67 and high Ki-67 groups and oligodendroglioma,IDH-mutant,and 1p/19q-codeleted WHO 2 and WHO 3 groups by ITSS were 0.662,0.636,0.794 and 0.827,respectively,with accuracy of 65.2%,55.4%,75%and 90.5%,sensitivity of 44.4%,69.6%,67.3%and 75%,specificity of 85.1%,50.7%,79.2%and 100%,and cut-off value of 1.5,2.5,2.5 and 1.5.There was no significant correlation between ITSS and grading of astrocytoma,IDH-mutant.Conclusion:SWI can be used for the prediction of IDH1 mutation status,1p/19q codeletion status,proliferation of gliomas and grading of oligodendroglioma,IDH-mutant,and 1p/19q-codeleted.Part Ⅵ The comparison and optimization of multiparametric functional MRI in the prediction of IDH1 mutation status,1p/19q codeletion status and proliferation of gliomasObjective:To compare and optimize the parameters of multiparametric functional MRI(including ASL、multi-b value DWI、DKI、APT and SWI)in the prediction of IDH1 mutation status,1p/19q codeletion status and proliferation of gliomas,and provide a simple functional MRI evaluation scheme.Methods:All the patients diagnosed with suspicious supratentorial adult-type diffuse gliomas underwent ASL,multi-b value DWI,DKI,APT and SWI before operation,and 92 patients were confirmed by histopathology.All the parameters were compared between IDH1 mutation and IDH1 widetype groups,1p/19q codeletion and 1p/19q non-codeletion groups and low Ki-67 and high Ki-67 groups using Mann-Whitney U test respectively,and P<0.05 was regarded as statistically significant.ROC curve was used to assess the diagnostic efficacy of statistically significant datas,and multivariate logistic regression analysis and the least absolute shrinkage and selection operator(LASSO)was used to select the best function parameters.Results:The combination of ASL,multi-b value DWI,DKI,APT and SWI demonstrated good performance for differentiating between IDH1 mutation and IDH1 widetype groups,and was better than single functional MRI,with AUC of 0.722,accuracy of 71.7%,sensitivity of 60%,and specificity of 83%.The AUCs of multi-b value DWI,SWI,DKI,APT and ASL were 0.702,0.662,0.658,0.658 and 0.65 respectively,and multi-b value DWI parameter-slow ADC and ASL parameter-CBF were the best function parameters in differentiating between IDH1 mutation and IDH1 widetype gliomas.The combination of all the functional MRI demonstrated good performance for differentiating between 1p/19q codeletion and 1p/19q non-codeletion groups,and was better than single functional MRI,with AUC of 0.79,accuracy of 69.6%,sensitivity of 87%,and specificity of 63.8%.The AUCs of multi-b DWI,ASL,DKI and SWI were 0.766,0.73,0.679 and 0.636 respectively.Slow ADC and fraction of fast ADC derived from multi-b value DWI and CBF derived from ASL were the best function parameters in differentiating between 1p/19q codeletion and 1p/19q non-codeletion gliomas.There was no significant correlation between APT and 1p/19q codeletion status of gliomas.The combination of all the functional MRI demonstrated good performance for differentiating between low Ki-67 and high Ki-67 groups,and was better than single functional MRI,with AUC of 0.901,accuracy of 85.9%,sensitivity of 81.6%,and specificity of 90.7%.The AUCs of multi-b value DWI,DKI,SWI,APT and ASL were 0.847,0.799,0.794,0.69 and 0.634 respectively.Slow ADC and fraction of fast ADC derived from multi-b value DWI and SWI were the best function parameters in differentiating between low Ki-67 and high Ki-67 gliomas.Conclusion:Multi-b value DWI(slow ADC and fraction of fast ADC)and ASL(CBF)were the optimized functional MRI sequences(parameters)in the prediction of IDH1 mutation status and 1p/19q codeletion status of gliomas;and multi-b value DWI(slow ADC and fraction of fast ADC)and SWI(ITSS)were the optimized functional MRI sequences(parameters)in the prediction of proliferation of gliomas.The optimization of multiparametric functional MRI can simplify the examination procedure,shorten the examination time and improve the diagnostic efficiency.Chapter Ⅱ The study of radiomics in the prediction of IDH1 mutation status,1p/19q codeletion status and proliferation of gliomasPart I The study of the value of multiparametric conventional MRI-based radiomics in the prediction of IDH1 mutation status,1p/19q codeletion status and proliferation of gliomasObjective:To evaluate the value of multiparametric conventional MRI-based radiomics in the prediction of IDH1 mutation status,1p/19q codeletion status and proliferation of gliomas.Methods:All the patients diagnosed with suspicious supratentorial adult-type diffuse gliomas underwent conventional MRI before operation,including T2WI,T2 FLAIR and CE T1WI,and 92 patients were confirmed by histopathology.During image processing,image registration was conducted.A radiomics analysis was performed by extracting multiparametric conventional MRI-based features,and the least absolute shrinkage and selection operator(LASSO)was applied to generate a radiomics signature for predicting IDH1 mutation status,1p/19q codeletion status and proliferation of gliomas.Statistical analysis included analysis of variance,Pearson correlation,multivariate logistic regression,Hosmer-Lemeshow test and ROC curve.Results:The radiomics signature demonstrated good performance for differentiating between IDH1 mutation and IDH1 widetype groups,lp/19q codeletion and 1p/19q non-codeletion groups,and low Ki-67 and high Ki-67 groups,with AUCs of 0.779,0.888 and 0.955,accuracy of 72.5%,85%and 91.2%,sensitivity of 77.5%,80%and 95.2%,and specificity of 67.5%,86.7%and 86.8%,respectively.Conclusion:This study indicated that multiparametric conventional MRI-based radiomics had high diagnostic efficacy in the preoperative prediction of IDH1 mutation status,1p/19q codeletion status and proliferation of gliomas,and multi-center studies with greater sample sizes are needed in the future.Part Ⅱ The study of the value and optimization of different radiomics in the prediction of IDH1 mutation status,1p/19q codeletion status and proliferation of gliomasObjective:To evaluate the value of optimized multiparametric functional MRI-based radiomics and different functional MRI combined conventional MRI-based radiomics in the prediction of IDH1 mutation status,1p/19q codeletion status and proliferation of gliomas,and select the best radiomics model.Methods:All the patients diagnosed with suspicious supratentorial adult-type diffuse gliomas underwent conventional MRI(including T2WI,T2 FLAIR and CE T1WI)and functional MRI(including multi-b value DWI,3D-ASL and SWI)before operation,and 92 patients were confirmed by histopathology.During image processing,image registration was conducted.A radiomics analysis was performed by extracting optimized functional MRI(including slow ADC,fraction of fast ADC,3D-ASL and SWI)-based features,3D-ASL combined conventional MRI-based features,optimized multi-b value DWI functional parameters(including slow ADC and fraction of fast ADC)combined conventional MRI-based features,SWI combined conventional MRI-based features,and optimized multiparametric functional MRI combined conventional MRI-based features,and the LASSO was applied to generate radiomics signatures for predicting IDH1 mutation status,1p/19q codeletion status and proliferation of gliomas.Statistical analysis included analysis of variance,Pearson correlation,multivariate logistic regression,Hosmer-Lemeshow test and ROC curve.Results:The optimized multiparametric functional MRI combined conventional MRI-based radiomics demonstrated good performance for differentiating between IDH1 mutation and IDH1 widetype groups,with AUC of 0.904,accuracy of 72.5%,sensitivity of 83.8%,and specificity of 95%,and was better than the conventional MRI-based radiomics and other combined MRI-based radiomics.The AUCs of the optimized multiparametric functional MRI-based radiomics,optimized multi-b value DWI functional parameters combined conventional MRI-based radiomics,ASL combined conventional MRI-based radiomics,SWI combined conventional MRI-based radiomics and conventional MRI-based radiomics were 0.899,0.891,0.826,0.819 and 0.799,respectively.The optimized multiparametric functional MRI combined conventional MRI-based radiomics and ASL combined conventional MRI-based radiomics demonstrated good performance for differentiating between 1p/19q codeletion and 1p/19q non-codeletion groups,with AUCs of 0.916,accuracy of 80%,sensitivity of 90%,and specificity of 76.7%,and was better than the conventional MRI-based radiomics and other combined MRI-based radiomics.The AUCs of the optimized multi-b value DWI functional parameters combined conventional MRI-based radiomics,SWI combined conventional MRI-based radiomics,conventional MRI-based radiomics and optimized multiparametric functional MRI-based radiomics signature were 0.914,0.904,0.888 and 0.864,respectively.The optimized multiparametric functional MRI combined conventional MRI-based radiomics and multi-b value DWI functional parameters combined conventional MRI-based radiomics demonstrated good performance for differentiating between low Ki-67 and high Ki-67 groups,with AUC of 0.961,accuracy of 90%,sensitivity of 88.1%,and specificity of 92.1%,and was better than the conventional MRI-based radiomics and other combined MRI-based radiomics.The AUCs of ASL combined conventional MRI-based radiomics,SWI combined conventional MRI-based radiomics,conventional MRI-based radiomics and the optimized multiparametric functional MRI-based radiomics were 0.955,0.955,0.955 and 0.888,respectively.Conclusion:This study indicated that optimized multiparametric functional MRI combined conventional MRI-based radiomics could improved the predictive ability of conventional MRI-based radiomics in IDH1 mutation status,1p/19q codeletion status and proliferation of gliomas,especially in genetic molecular diagnosis.It is helpful for preoperative diagnosis of gliomas,determination of treatment plan and evaluation of prognosis,but multi-center studies with greater sample sizes are needed in the future.