Radiomics Based On Multimodal MRI For The Differentiation Of The Recurrence And Treatment Response Of Glioma

ObjectiveTo explore the application value of radiomics and nomogram based on Contrast-enhanced T1 Weighted Imaging(CE-T1WI),Cerebral Blood Flow(CBF)map,Amide Proton Transfer Weighted(APTW)imaging,Quantiative Susceptibility Mapping(QSM)and Apparent Diffusion Coefficient(ADC)map in differentiating the recurrence and treatment response of glioma.Materials and MethodsThe clinical datas and imaging datas of 96 patients who developed abnormal enhancement after glioma surgery were retrospectively analysed.Clinical factors were compared between the two groups using univariate analysis,and clinical model were developed using logistic regression model.Patients were randomly divided into training group and validation group according to 7:3.Region of Interest(ROI)were delineated in CE-T1WI,CBF map,APTW,QSM and ADC map and radiomics features were extracted using 3D-Slicer software.The training group was used to screen the features,develop the radiomics model and nomogram.The validation group was used to further verify the performance of the model.Features screening was performed by Student’s t test,Mann-Whitney U test and LASSO regression algorithm.Five single-sequence models and one combined model based on CE-T1WI,CBF map,APTW,QSM and ADC map were developed.The Radiomics score(Rad-score)was calculated for each patient based on the features of the combined model and combined with the Rad-score and IDH1 genotype to develop a nomogram.The diagnostic performance of each model was evaluated using the Receiver Operator Characteristic(ROC)curve and the Area Under the Curve(AUC),and the sensitivity,specificity and accuracy of the models were calculated.The Delong test was used to compare the AUC of the models,and P<0.05 was considered statistically significant.Calibration curve,the Hosmer-Lemeshow test and Decision Curve Analysis(DCA)were used to assess the calibration,goodness of fit and clinical value of the nomogram.Results1.After univariate analysis,age,maximum tumour diameter and IDH1 genotype could be used to develop the clinical model,which had the ability to distinguish recurrence from treatment response,with an AUC of 0.738 in the training group and 0.701 in the validation group.Multivariate analysis showed that IDH1 genotype in the two groups were statistically significant(P<0.05),and IDH1 genotype were selected from clinical factors to develop nomogram.2.The five single-sequence radiomics models can distinguish the two groups,but the diagnostic performance is moderate.The AUC of the training group is 0.761-0.867,and the AUC of the validation group is 0.708-0.757.The combined model based on multimodal MRI(CBF map,APTW,QSM and ADC map)had a high diagnostic performance,with an AUC of 0.956 in the training group and 0.951 in the validation group.The Delong test showed that the difference in AUC between the combined model and the five single-sequence radiomics models was statistically significant(P<0.05).3.The AUC of the nomogram was 0.975 in the training group and 0.972 in the validation group.According to Delong test,there was no significant difference in AUC between the nomogram and the combined model(P>0.05),and the diagnostic performance of the nomogram and the combined model was similar,and both of them could well distinguish recurrence from treatment response.The calibration curve and Hosmer-Lemeshow test showed good agreement and goodness of fit for the nomogram.The DCA curve showed that the nomogram had the highest net benefit within a wide range of threshold probabilities(0-91%)and had the highest clinical value.ConclusionCompared with the single-sequence radiomics model,the combined model combined with CBF map,APTW,QSM and ADC map can improve the discrimination ability between recurrence and treatment response.Both the combined model and the nomogram can be used to distinguish the recurrence and treatment response of glioma,and both have high diagnostic performance.But the nomogram has higher clinical value.