The Application Of Diffusion Kurtosis Imaging And Histogram Analysis In Staging Of Nasopharyngeal Carcinoma And Assessment Of Treatment Response Of Radiotherapy And Chemotherapy

Part oneObjective To evaluate the correlation between histogram parameters derived from diffusion kurtosis imaging(DKI)and the TNM staging and clinical staging of nasopharyngeal carcinoma(NPC).Materials and methods Forty-seven patients with initial diagnosis of locally advanced NPC and DKI acquisitions before treatment were retrospectively enrolled.Apparent diffusion coefficient(ADC)was calculated based on monoexpotential model.Apparent diffusion kurtosis coefficient corrected by non-Gaussian distribution(D)and apparent kurtosis coefficient(K)were calculated based on diffusion kurtosis model.Histogram parameters(10th,25th,50th,75th,90th percentile,mean,standard deviation,skewness,and kurtosis)of DKI-derived parameters(ADC,D,K)were calculated.Patients were divided into low and high T,N,M and clinical stage based on American Joint Committee on Cancer(AJCC)staging system(8th edition).Intraclass correlation coefficient(ICC),Kolmogorov-Smimov test,Student's t-test or Mann-Whitney U test,receiver operating characteristic curve(ROC)and Delong tests were performed.Results Most of the parameters have good to excellent consistency(ICC:0.885-0.998).Compared to low T stage patients,NPC patients with high T stage have significantly higher ADCkur(P = 0.036)and Dkur(P = 0.043).Compared to low N stage patients,NPC patients with high N stage have significantly lower ADCsd(P=0.025).ROC analysis indicated that setting ADCkur =4.75 as the cut-off value could obtain optimal diagnostic performance for the prediction of T stage,while ADCsd=0.145 ×10-3 mm2/s is optimal for prediction of N stage.The DKI parameters showed no significant difference between different M stage and AJCC stage(all P>0.05).Conclusion DKI derived parameters may be used as promising quantitative biomarkers in prediction of T staging and N staging of NPC.Part twoObjective To explore the quantitative indices for chenmoradiotherapy treatment response assessment of nasopharyngeal carcinoma(NPC)based on histogram analysis of diffusion kurtosis imaging(DKI).Materials and methods Thirty-six patients with initial diagnosis of locally advanced NPC and DKI acquisitions before and after neoadjuvant chemotherapy(NAC)were enrolled.Patients were divided into respond(RP)and non-respond(NRP)groups and residual(RD)and non-residual(NRD)groups according to NAC and radiation therapy(RT)response.Histogram parameter(10th,25th,50th,75th,90th percentile,mean,standard deviation,skewness,and kurtosis)of DKI-derived parameters(ADC,D,K)were calculated.Intraclass correlation coefficient(ICC),Kolmogorov-Smirnov test,Student's t-test or Mann-Whitney U test,receiver operating characteristic curve(ROC)and Delong tests were performed.Results Most of the parameters have good to excellent consistency(ICC:0.675-0.998).The pre-and post-ADC(10th,25th,50th),D(10th,25th,50th)and K(50th)were significantly different between RP and NRP,while the pre-and post-ADC(10th,50th),D.(25th,50th)and K(75th,90th)were significantly different between RD and NRD(all P<0.05).ROC analysis indicated that setting pre-D50th=0.875×10-3mm2/s as the cut-off value could obtain optimal diagnostic performance for the prediction of NAC treatment response,while post-K90th=1.035 is optimal for prediction of RT response.Conclusion DKI derived parameters can be used as promising quantitative biomarkers in prediction of NAC and RT treatment response in locally advanced NPC patients.Part threeObjective To explore the effect of seletion of different regions of interest(ROI)in chemoradiotherapy treatment response assessment of nasopharyngeal carcinoma(NPC)based on histogram analysis of diffusion kurtosis imaging(DKI).Materials and methods Thirty-six patients with initial diagnosis of locally advanced NPC and DKI acquisitions before and after neoadjuvant chemotherapy(NAC)were enrolled.Patients were divided into respond(RP)and non-respond(NRP)groups and residual(RD)and non-residual(NRD)groups according to NAC and radiation therapy(RT)response.Based on previous study,ADC(10th,50th),D(10th,25th,50th)and K(50th)were calculated by two radiologists using three different ROI drawing methods.Intraclass correlation coefficient(ICC),Kolmogorov-Smirnov test,Student's t-test or Mann-Whitney U test,receiver operating characteristic curve(ROC)and Delong tests were performed.Results Most of the parameters have good to excellent consistency(ICC:0.909-0.998).ADC50th and D25th calculated using whole tumor volume ROI(WTV)is significantly different with that using single section ROI(SS)and three sections ROI(3S)(P<0.05).The three methods had significant difference when calculating D50th hand K50th.The D10th-SS,ADC(10th,50th)-WTV,D(10th,25th,50th)-WTV and K(50th)-WTV were significantly different between RP and NRP,while ADC(10th)-SS,D(50th)-SS,ADC(10th,50th)-3S,D(50th)-3S,ADC(10th,50th)-WTV,D(10th,25th,50th)-WTV and K(50th)-WTV were significantly different between RD and NRD.ROC analysis indicated that setting D50th-WTV=0.875×10-3mm2/s as the cut-off value could obtain optimal diagnostic performance for the prediction of NAC treatment response,while ADC10th-WTV= 0.515×10-3mm2/s is optimal for prediction of RT response.Conclusion Whole tumor volume ROI should be preferred in quantitative treatment response assessment of nasopharyngeal carcinoma(NPC)based on diffusion kurtosis imaging(DKI).