Diagnostic Performance Of The Multimodal Magnetic Resonance Imaging In Preoperative Glioma Grading

Objective1.Using multimodal magnetic resonance imaging(conventional and contrast-enhanced,DKI,IVIM,DSC),and analyzing the tumour lesion area,the peritumoral edema area,and the contralateral normal appearing white matter area of brain glioma.The value of MRI parameters in the diagnosis of glioma grading and the correlation between MRI parameters and pathological grade were preliminarily discussed.2.By analyzing the tumour lesion area and pathological grade of brain glioma,the optimal parameters of multimodal magnetic resonance imaging for the diagnosis of glioma grade were discussed.3.To evaluate the diagnostic efficacy of multimodal magnetic resonance imaging for glioma grading,and to provide an objective basis for the individualized diagnosis and treatment of clinical patients.Materials and Methods1.Seventy-one consecutive patients(range18-77 years of age;mean,45.6±11.1 years,37 males(mean age,44.1±9.8years),34 females(mean age,47.2±12.4years);30 low-grade gliomas(LGG,grades I [n = 2] and II [n = 28])and 41 high-grade gliomas(HGG,grades III[n = 25] and IV [n = 16])with histopathologically confirmed glioma(All patients were not treated with radiotherapy,chemotherapy or other anti-tumor therapy before operation)who provided written informed consent were retrospectively assessed between 09/2015 and 11/2017 from a prospective trial approved by the local institutional review board.2.Patients were examined with a 1.5T scanner(Signa Excite HDx;GE Healthcare,Milwaukee,WI,USA)using an 8-channel phased-array head coil.All patients underwent conventional MR imaging,DKI,IVIM and DSC.Parameters of DKI(fractional anisotropy(FA),mean diffusion(MD),axial diffusion(Da),radial diffusion(Dr),fractional anisotropy kurtosis(FAK),mean kurtosis(MK),axial kurtosis(Ka)and radial kurtosis(Kr)),IVIM(standard apparent diffusion coefficient(sADC),slow diffusion coefficient(D),fast diffusion coefficient(D*),fraction of fast ADC(f),distributed diffusion coefficient(DDC)and water molecular diffusion heterogeneity index(?))and DSC(relative cerebral blood flow(rCBF),relative cerebral blood volume(rCBV),mean transmit time(MTT)and time to peak(TTP))were generated and measured from tumor lesion areas(L),peripheral edema(E),contralateral normal appearing white matter(CNAWM),and then normalized(lesion areas/ contralateral normal appearing white matter)each parameter of the lesion area.All multimodal MR data were analyzed and processed on an ADW4.6 workstation(Function tool;GE Healthcare,Milwaukee,WI,USA).3.Data were expressed as mean ± standard deviation.Statistical analysis was performed using SPSS v.13.0 software(SPSS Inc.,Chicago,IL,USA).P values < 0.05 were considered statistically significant for all tests.One-way analysis of variance was used to compare the normalized lesion area,edema,and CNAWM values of DKI(FA,MD,Da,Dr,FAK,MK,Ka,and Kr),IVIM(sADC,D,D*,f,DDC,and ?)and DSC(rCBF,rCBV,MTT,and TTP)metrics between LGG and HGG.A receiver operating characteristic(ROC)curve analysis was performed to determine optimal thresholds for glioma grading by each of the normalized parameters.In addition,sensitivity,specificity,and area under the curve(AUC)for glioma grading were calculated in each case.Relationships between each normalized lesion area parameter and tumor grade were analyzed with Spearman's correlation.A binary logistic regression analysis was conducted to evaluate the diagnostic accuracy of single-mode MR(DKI,IVIM and DSC)and multimodal MR in glioma grading.Results1.DKIIn the tumor lesion area,mean values of normalized MK,Ka and Kr were lower for LGG than for HGG(P < 0.05),whereas the opposite trend was observed for mean values of normalized MD,Da,Dr and FAK(P < 0.05).There was no difference in the FA value of the lesion area between LGG and HGG(P > 0.05).The mean Da value was lower for LGG than for HGG(P < 0.05)in E.MD and Da values were lower in LGG than in HGG(P < 0.05),whereas MK and Ka values in the CNAWM were higher(P < 0.05).The mean values of FA,MK,Ka and Kr in tumor lesion area were lower than in the CNAWM,and the mean values of MD,Da and Dr in tumor lesion area were higher than in the CNAWM.2.IVIMMean values of normalized sADC,D,D*,DDC and ? were higher in LGG than in HGG(P < 0.05)in the lesion area,whereas no differences were observed in the f value of the lesion area or in any of the IVIM parameters of E between LGG and HGG(P > 0.05).Mean values of DDC and ? were slightly lower in LGG than in HGG in the CNAWM(P < 0.05).The mean values of sADC,D,D*,DDC and ? in tumor lesion area were lower than in the CNAWM,and the mean values of f in tumor lesion area were higher than in the CNAWM.3.DSCMean values of normalized rCBF and rCBV were lower for LGG than for HGG in the lesion area(P< 0.05).However,mean normalized MTT and TTP values of the lesion area did not differ between LGG and HGG(P>0.05).The mean TTP value was lower for LGG than for HGG(P < 0.05)in E.There were no statistical differences in rCBF,rCBV,MTT and TTP in the CNAWM between LGG and HGG(P > 0.05).The mean values of rCBF and rCBV in tumor lesion area were higher than peripheral edema area and in the CNAWM.4.An ROC analysis was carried out to determine the optimal thresholds,AUC,sensitivity,and specificity for differentiating the performance of normalized DSC,DKI,and IVIM DWI parameters in terms of diagnosing LGG and HGG.Normalized mean rCBV(2.240ml/100 g,0.849,87.8%,76.7%),MK(0.471,0.842,92.7%,76.7%)and ?(1.064,0.819,76.7%,78.0%)showed the best diagnostic performance for identifying glioma grade.5.There was a significant positive correlation between normalized rCBF,rCBV,MK,Ka,Kr,and tumor grade(P < 0.01),whereas negative correlations were observed between MD,Da,Dr,FAK,sADC,D,D*,DDC,and ? and tumor grade(P < 0.01).6.When we were combined all of the multimodal MR parameters,IVIM DWI(87.3%)showed the best diagnostic accuracy relative to other modalities,followed by DSC(81.7%)and DKI(80.3%).The diagnostic accuracy was 91.7% for the combination of DKI and IVIM,followed by DSC and IVIM(87.3%)and DSC and DKI(85.9%).The combination of three advanced MR modes(DSC,DKI,and IVIM)showed the highest accuracy for predicting tumor grade,with a diagnostic accuracy of 100%.Conclusion1.Multimodality magnetic resonance imaging is an effective imaging technique for the diagnosis of glioma in the preoperative period,and they complement each other.2.There was a significant positive correlation between normalized rCBF,rCBV,MK,Ka,Kr,and tumor grade,whereas negative correlations were observed between MD,Da,Dr,FAK,sADC,D,D*,DDC,and ? and tumor grade.3.Among the parameters of DKI,IVIM and DSC,the parameters of MK,? and rCBV are the best diagnostic for glioma grading,respectively.4.The single mode MR check was the highest accuracy of IVIM prediction,followed by DSC and DKI.It can better improve the accuracy of high and low grade glioma and provide help for patients' diagnosis and treatment.