Value Of Radiomics Based On Magnetic Resonance Imaging In Predicting Radiation-induced Temporal Lobe Injury In Nasopharyngeal Carcinoma

Part 1 A MRI-based radiomics model predicting radiation-induced temporal lobe injury in nasopharyngeal carcinomaPurpose:To develop and validate a radiomics-based model for predicting radiationinduced temporal lobe injury(RTLI)in nasopharyngeal carcinoma(NPC)by pretreatment MRI of the temporal lobe.Materials and Methods:A total of 216 patients with diagnosed NPC were retrospectively reviewed.Patients were randomly allocated to the training(n=136)and the validation cohort(n=80).Radiomics features were extracted from pretreatment contrast-enhanced T1-or fat-suppressed T2 weighted MRI.A radiomics signature was generated by the least absolute shrinkage and selection operator(LASSO)regression algorithm,Pearson correlation analysis,and multivariable Logistic analysis.Clinical features were selected with Logistic regression analysis.Multivariable Logistic regression analysis was conducted to develop three models for RTLI prediction in the training cohort:namely radiomics signature,clinical variables,and clinical-radiomics parameters.A radiomics nomogram was used and assessed with respect to calibration,discrimination,reclassification,and clinical application.Results:The radiomics signature,composed of two radiomics features,was significantly associated with RTLI.The proposed radiomics model demonstrated favorable discrimination in both the training(AUC,0.89)and the validation cohort(AUC,0.92),outperforming the clinical prediction model(P<0.05).The AUCs of the clinicalradiomics model for predicting RTLI in the training and validation cohorts were 0.93(95%CI:0.88,0.97)and 0.95(95%CI:0.90,1.00),respectively.The AUC value of the clinical-radiomics model was higher than that of the radiomics model.The clinicalradiomics nomogram demonstrated good agreement between predicted and observed RTLI in both the training and validation cohorts.The clinical-radiomics model showed also excellent performance in predicting RTLI in different clinical-pathologic subgroups.Conclusions:A radiomics model derived from pretreatment MRI of the temporal lobe as imaging biomarkers have the potential to prediction radiation-induced temporal lobe injury in nasopharyngeal carcinoma.Part 2 Added value of histogram analysis of ADC in predicting radiation-induced temporal lobe injury of patients with nasopharyngeal carcinoma treated by intensity-modulated radiotherapyPurpose:To evaluate the predictive potential of histogram analysis derived from apparent diffusion coefficient(ADC)maps in radiation-induced temporal lobe injury(RTLI)of nasopharyngeal carcinoma(NPC)after intensity-modulated radiotherapy(IMRT).Materials and Methods:Pretreatment DWI of temporal lobes of 214 patients with NPC was retrospectively included to obtain ADC histogram parameters.Patients were randomly divided into training and validation sets(135:79).Univariate and multivariate analyses were used to identify the significant histogram parameters and clinic-dosimetric variables in the training cohort,which was then incorporated into a combined predictive model.The area under the curve(AUC)of the receiver operating characteristic curve was used to evaluate the performance of significant variables and combinations.A nomogram was used and assessed concerning calibration,discrimination,reclassification,and clinical application.Results:Kurtosis(P=0.06),maximum(P=0.05),range(P=0.06),and total-energy(P<0.001)were significant predictors of RTLI.A Rad-score was established combing those four significant variables,and it provides an AUC of 0.95(95%CI:0.91,0.98)and 0.89(95%CI:0.81,0.97)in the training and validation cohort,respectively.The combined model,integrating Rad-score with T stage(P=0.02)showed a favorable prediction performance in the training and validation cohorts(AUC=0.96 and 0.87).Calibration curves determined a good agreement between the predicted and actual RTLI occurrence.Conclusions:Pretreatment ADC histogram and its combination with the T stage had satisfactory ability to predict RTLI in NPC after IMRT.Part 3 MRI-based radiomics model for predicting radiation-induced temporal lobe injury in nasopharyngeal carcinoma after intensity-modulated radiotherapyPurpose:To develop a model based on magnetic resonance imaging(MRI)radiomics and clinical features for predicting radiation-induced temporal lobe injury(RTLI)in patients with nasopharyngeal carcinoma(NPC)after intensity-modulated radiotherapy(IMRT).Materials and Methods:Two hundred and sixteen patients with NPC were retrospectively included.MR images of patients within 1 week after radiotherapy were analyzed.Radiomics features were extracted and selected based on post-treatment temporal lobe MR images.Clinical variables were selected by multivariate Logistic regression analysis.The Logistic regression analysis was performed for RTLI prediction models construction.The area under the receiver operating characteristic curve(AUC)was calculated for performance evaluation.Results:Three radiomics features(CET1_wavelet-HHH_glszm_SmallAreaEmphasis,T2_lbp-3D-m1_firstorder_Kurtosis,T2_lbp-3Dk_ngtdm Coarseness)were selected to construct the radiomics signature(AUC of 0.94 and 0.92).The clinical-radiomics model,integrating radiomics signature with T classification,achieved higher predictive performance in the training and validation cohorts(AUC of 0.95 and 0.93),as well as improved accuracy of the classification of RTLI outcomes(net reclassification improvement:0.711;95%CI:0.57,0.86;P<0.001).Conclusions:The clinical-radiomics model and radiomics signature both showed great performance in predicting RTLI in patients with NPC.