Application of neurite orientation dispersion and density imaging to predict IDH genotype of adult diffuse glioma
Objective:To evaluate the differences of tumors with different isocitrate dehydrogenase(IDH)gene states by using the quantitative parameters of neurite orientation dispersion and density imaging(NOODI)and to explore the clinical application value of NOODI in genotyping of adult-type diffuse glioma(ADG).Materials and Methods:Fifty-one patients with adult diffuse glioma(IDH mutant 21 cases,IDH wild type 30 cases)were collected.Routine scanning and multi-shell diffusion scanning were performed before operation,and four quantitative parameter maps were obtained after image preprocessing and post-processing.All patients underwent conventional scan and multi-shell diffusion scan before operation.Four quantitative parametric maps were obtained after image preprocessing and post-processing.The 3D-Slicer was used to delineate the regions of interest in tumor solid areas,peritumoral edema areas,and contralateral normal white matter areas,compared the differences between groups with different IDH status.FeAture Explorer(FAE)software was used to construct the prediction model,and multiple pipeline combinations were considered during the development of the prediction model,including 1 data dimensionality reduction method(Pearson's correlation coefficient),4 feature selection methods(multivariate analysis of variance,Kruskal Wallis test,recursive feature elimination and Relief algorithm)and 4 linear classifiers(support vector machine,linear discriminant analysis,logistic regression and logistic regression with LASSO regularization term),with a total of 16 pipelines.The comparisons between models were using the receiver operating characteristic(ROC)curve.The final model was evaluated by integrated discrimination improvement(IDI),net reclssification improvement(NRI)index and leave-one-out cross-validation.Results:The age of IDH wild-type group was higher than that of IDH mutant group(P=0.001).Only the orientation dispersion index(ODI)of IDH wild-type group was lower than that of IDH mutant group(P=0.019),and there was no difference among other parameter groups.The contralateral white matter area and edema area had the highest and lowest intra-cellular volume fraction(ICVF)values,respectively(P<0.05).The comprehensive model constructed by linear discriminant analysis has the highest diagnostic efficiency,and the area under ROC curve is 0.835(95%CI:0.703-0.941).It is higher than the age of single use[0.773(95%CI:0.624-0.894)]and ODI[0.695(95%CI:0.538-0.845)].The IDI and NRI prove that the final model has the highest diagnostic performance(all P<0.001).Decision curve analysis and calibration curve prove the clinical net benefit of the final model and the distribution closest to the real data(Brier score is 0.163).Conclusions:The quantitative parameters of NODDI can be used to describe the microenvironment differences of ADG.The final model can effectively predict the different IDH gene states of ADG,and the diagnostic efficiency of the composite model is better than that of the single model.
gliomagenotypingisocitrate dehydrogenaseneurite directional dispersion and density imagingmagnetic resonance imagingbiomark
国家科技期刊平台
NSTL
万方数据
维普
