| Objective:Diffusion kurtosis imaging?DKI?,as a new technology,can quantify the non-Gaussian diffusion behavior of water molecules in tissue,which has attracted more and more attention.Dynamic contrast-enhanced magnetic resonance imaging?DCE-MRI?can reflect the physiological activity of microcirculation through imaging data.It also can provide hemodynamic information.There are many kinds of tumors in the head and neck region,and distinguishing different types of tumors plays a decisive role in the choice of clinical treatment.The purpose of this study was to explore the diagnostic value of DKI and DCE-MRI in distinguishing the types of head and neck tumors.Methods:79 head and neck tumors were analyzed retrospectively,including 44 benign lesions and 35 malignant lesions,which were underwent DKI and DCE-MRI scan.The differences of mean diffusivity?MD?,mean kurtosis?MK?,time-signal intensity curve?TIC?type,peak time?Tpeak?,and washout ratio?WR?between benign and malignant head and neck lesions were compared.At the same time,we compared the diagnostic accuracy of DKI alone or DCE-MRI alone and combination of the two.The head and neck tumors were further divided into four subgroups,including benign tumor group?excluding adenolymphoma?,malignant tumor group?excluding lymphoma?,lymphoma group and adenolymphoma group,and compare the differences of parameters of DKI and DCE-MRI in four subgroups.The parameters of DKI and DCE-MRI of the four groups were compared by stepwise analysis.The parameters of benign tumor group and?Lymphoma group+adenolymphoma group+malignant tumor group?,malignant tumor group and?Lymphoma group+adenolymphoma group?,lymphoma group and adenolymphoma group were analyzed by the receiver operating characteristic curve?ROC?.The area under the curve?AUC?was obtained,and the best diagnostic threshold,sensitivity,specificity,and accuracy were calculated to compare the diagnostic ability of each parameter.Results:MD and Tpeak of benign lesions were significantly higher than those of malignant lesions?P?0.009?,and MK of benign lesions was lower than that of malignant lesions?P<0.001?.MD=1.44×10-3mm2/s,MK=0.80 and Tpeak=86.80s were used as the best diagnostic thresholds for malignant lesions,with the AUC of 0.822,0.812 and 0.670respectively,the sensitivity of 100%,88.57%,and 100%respectively,the specificity of 56.82%,68.18%,and 40.91%respectively,the accuracy of75.95%,77.22%,and 67.09%respectively.The diagnostic ability of MD and MK was both better than that of Tpeak.There were significant differences in TIC types between benign and malignant lesions?P<0.001?.Combination use of DKI and DCE-MRI indicated that overall A-type lesions were benign lesions.Benign lesions showed higher WR than malignant lesions of B-type TIC.WR=45.00%was used as the best diagnostic threshold,with the AUC of 0.967,the sensitivity of 92.31%,the specificity of 93.33%,and the accuracy of 92.86%.Benign lesions showed higher MD than malignant lesions of C-type TIC.MD=1.22×10-3mm2/s was used as the best diagnostic threshold,with the AUC of0.958,the sensitivity of 81.82%,the specificity of 100%,and the accuracy of 88.89%.Benign lesions showed lower MK than malignant lesions of C-type TIC.MK=0.74 was used as the best diagnostic threshold,with the AUC of 0.976,the sensitivity of 86.36%,the specificity of 100%,and the accuracy of 91.67%.In the further comparison of the four subgroups,there were significant differences in the comparison of MD,MK,Tpeak,and WR in each subgroup?P<0.001?.There were significant differences in MK and WR between lymphoma group and adenolymphoma group,in MD and MK between lymphoma group and malignant tumor group,and in MK and WR between adenolymphoma group and malignant tumor group.MD,MK,Tpeak,and WR of benign tumor group were significantly different from the other three groups in a pairwise comparison.A stepwise analysis indicated that benign tumor group showed higher MD?P<0.001?,higher Tpeak?P<0.001?,lower MK?P<0.001?,and lower WR?P<0.001?than?Lymphoma group+adenolymphoma group+malignant tumor group?.MD=1.32×10-3mm2/s,MK=0.80,Tpeak=68.20s,and WR=16.13%were used as the best diagnostic thresholds respectively,with the AUC of0.969,0.962,0.854,0.877 respectively,the sensitivity of 87.50%,93.75%,75.00%,87.50%respectively,the specificity of 93.62%,91.49%,87.23%,80.85%respectively and the accuracy of 91.14%,92.41%,82.28%,83.54%respectively.Malignant tumor group showed higher MD and Tpeak?P?0.004?,lower MK and WR?P?0.001?than?Lymphoma group+adenolymphoma group?.MD=1.04×10-3mm2/s,MK=1.10,Tpeak=24.80s,and WR=51.16%were used as the best diagnostic thresholds respectively,with the AUC of 0.941,0.924,0.734,0.782 respectively,the sensitivity of 80.95%,85.71%,76.19%,100%respectively,the specificity of 100%,84.62%,69.23%,46.15%respectively and the accuracy of 91.49%,85.11%,72.34%,70.21%respectively.Adenolymphoma group showed higher MD and WR?P<0.001?,and lower MK?P<0.001?than lymphoma group.MD=0.76×10-3mm2/s,MK=1.22,and WR=41.35%were used as the best diagnostic thresholds respectively,with the AUC of 0.985,0.875,1.000 respectively,the sensitivity of 100%,75.00%,100%respectively,the specificity of85.71%,92.86%,100%respectively and the accuracy of 92.31%,84.62%,100%respectively.Conclusion:There were significant differences in MD,MK,and Tpeak between the benign and malignant lesions of the head and neck,but there were overlaps among different tumor subtypes.A-type TIC of DCE-MRI indicates benign lesions.Compared with the use of DKI or DCE-MRI alone,combination of DKI and DCE-MRI has higher diagnostic accuracy in distinguishing lesions.The MK value of lymphoma group,adenolymphoma group,malignant tumor group,and benign tumor group decreased gradually.Compared with MD,Tpeak and WR,MK has higher sensitivity in distinguishing head and neck tumors.High MD,low MK,long Tpeak,and low WR often indicated benign tumor group.High MK and markedly high WR indicated adenolymphoma group.Low MD and markedly high MK indicated lymphpma group.Four groups of head and neck tumors can be effectively identified through the stepwise analysis of the parameters of DKI and DCE-MRI. |
PDF Full Text Request