| Background and PurposeParotid tumors have a wide variety of pathologic type,of which about 68%are benign tumors(BTs).The prevalent BTs include pleomorphic adenoma(PA),Warthin tumor(WT)and basal cell adenoma(BCA).The most common malignant tumor(MT)is mucoepidermoid carcinoma,followed by adenoid cystic carcinoma.The clinical manifestations of parotid tumors are similar,but the treatment strategies and prognosis usually are of markable difference.Therefore,accurate preoperative diagnosis is of great significance.Magnetic resonance imaging(MRI)was considered to be the most sensitive and specific imaging method in the diagnosis of parotid tumors.The apparent diffusion coefficient(ADC)value of diffusion weighted imaging(DWI)could quantitatively evaluate the Brownian motion diffusion of water molecules in tissues,which is of great significance in the diagnosis and identification of parotid tumors.The recently emerged readout segmentation of long variable echo-trains DWI(RESOLVE-DWI)was proved to reveal fewer artifacts,less deformation and better quality than conventional DWI.Moreover,Diffusion kurtosis imaging(DKI)has provided an advanced model to quantify the non-Gaussian behavior of water molecule diffusion.There were few reports on the identification of parotid tumors with RESOLVE-DKI.T2 mapping is a quantitative MR imaging modality,which could improve the accuracy of diagnosis by measuring T2 value of the lesions.Previous reports on the application of T2 mapping in parotid tumors are few.Similarly,T1 mapping is a newborn MRI technology,and T1 value could objectively reflect the biological tissue characteristics.However,its application value in parotid tumors has not been reported.Dynamic contrast enhanced MRI(DCE-MRI)is performed by a continuous dynamic scan using fast T1 weighted imaging(T1WI)after intravenous injection of contrast agent,and the time signal intensity curve(TIC)could be obtained.Previous studies have suggested that DCE-MRI could be used in the differential diagnosis of parotid tumors,however,there are defects such as low temporal resolution.The purpose of this study was to investigate the value of demographics,conventional MRI,RESOLVE-DWI,RESOLVE-DKI,T2 mapping,T1 mapping and DCE-MRI in the differential diagnosis of parotid tumors.Furthermore,this study established a combined diagnostic model to predict the pathologic type of parotid tumors,so as to improve the diagnostic accuracy.Materials and Methods1.Study Subjects and MRI Examination:The MRI data of 168 patients with parotid tumors confirmed by pathology from June 2018 to July 2021 in our hospital were collected.All the 168 patients underwent conventional MRI,RESOLVE-DWI,RESOLVE-DKI,T2 mapping,and T1 mapping scans before enhancement.Among them,133 patients underwent DCE and delayed enhanced scan,and 128 patients underwent T1 mapping scan after DCE scan.First,the parotid tumors were divided into BT and MT groups,then the BTs were further divided into PA group,WT group,and BCA group.2.Image,data processing and analysis:The demographic data and conventional MRI features of parotid tumors were analyzed and compared.ADC,K,D,T2,pre-enhanced T1(Tlpre),post-enhanced T1(Tlpost),Ktrans,Kep Ve,relative ADC(rADC),relative K(rK),relative D(rD),relative T2(rT2),relative T1pre(rTlpre),relative Tlpost(rT1post),absolute difference of T1 before and after enhancement(ΔT1),relative difference(change rate of T1 value),relative(rKtrans),relative Kep(rKep)and relative Ve(rVe)values of parotid tumors were measured,computed and compared.The TIC of parotid tumors were compared.The differential diagnostic efficacy of the above parameter values and TIC in parotid tumors were analyzed by the receiver operating characteristic(ROC)analysis.The sensitivity(SE),specificity(SP)were calculated.3.Building combined diagnostic models:The combined diagnostic model was constructed by logistic regression analysis,which was based on the parameters with diagnostic value obtained from the comparison of demographics and multi-modal MRI variables.Here,the parameter level of any two types of parotid tumor were used as the independent variable and the type of parotid tumor events in the research objects as the dependent variable.According to the formula PRE=1/(1+EXP(-Logit(P))),the predicted probability(P)of the subject suffering from a certain parotid tumor was obtained.When the PRE was greater than 0.5,the subject was predicted to be a certain parotid tumor type.Results1.Demographic characteristics of parotid tumors:The average age of patients with MT was higher than that of patients with BT,but the difference in gender was not significant(P=0.863).PA was more common in middle-aged women,and WT was more common in elder men.2.Conventional MRI findings of parotid tumors:MRI features of BT include well-defined margin,regular morphology,capsules,superficial lobes,and hypointensity on T1WI.PA usually displays hyperintensity on T2 weighted imaging(T2WI),with marked enhancement.WT may occur bilaterally,with low signal intensity on T2WI,cystic components,and low enhancement on contrast-enhanced T1WI.BCA usually displays relatively low signal on T2WI and marked enhancement.MT generally grows in an infiltrative fashion,with unclear borders,heterogeneous signals,hypotensity on T2WI,and moderate enhancement.3.RESOLVE-DWI analysis of parotid tumors:①The ADC value and rADC value of MT were lower than those of BT,and the differences were significant(all P<0.001).PA had the highest ADC value and rADC value,followed by BCA,and MT,WT was the lowest.② The AUC of ADC value and rADC value for differential diagnosis of BT and MT was 0.719 and 0.695,respectively.Taking the ADC value of 1.04×10-3mm2/s and rADC value of 0.72 as threshold values(TVs),the SE was 67.80%and 49.15%,and the SP was 74.00%and 84.00%,respectively.The AUC of ADC value and rADC value for differential diagnosis of PA and WT was 0.999 and 0.950,respectively.Taking the ADC value of 0.92×10-3mm2/s and rADC value of 0.28 as the TVs,the SE was 100.00%and 95.89%,and the SP was 96.55%and 82.76%,respectively.The AUC of ADC value for differential diagnosis of PA and BCA was 0.723.Taking the ADC value of 1.37×10-3mm2/s as the TV,the SE was 47.95%,and the SP was 90.91%.The AUC of ADC value and rADC value for differential diagnosis of PA and MT was 0.866 and 0.799,respectively.Using the ADC value of 1.04×10-3mm2/s and rADC value of 0.39 as the TVs,the SE was 93.15%and 86.30%,and the SP was 74.00%and 60.00%,respectively.The AUC of ADC value and rADC value for differential diagnosis of WT and BCA was 0.966 and 0.950,respectively.Using the ADC value of 0.80×10-3mm2/s and rADC value of 0.25 as the TVs,the SE was 100.00%,and the SP was 82.76%and 79.31%,respectively.The AUC of ADC value and rADC value for differential diagnosis of WT and MT was 0.694 and 0.646,respectively.Taking the ADC value of 0.80×10-3mm2/s and rADC value of 0.25 as the TVs,the SE was 82.76%and 79.31%,and the SP was 64.00%and 56.00%,respectively.The AUC of rADC value for differential diagnosis of BCA and MT was 0.795.Taking the rADC value of 0.25 as the TV,the SE was 100.00%,and the SP was 48.00%.4.RESOLVE-DKI analysis of parotid tumors:①The K value and rK value of MT were significantly higher than those of BT,and the D value and rD value of MT were significantly lower than those of BT,and the differences were significant(all P<0.05).WT had the highest K value and rK value,followed by MT and BCA,PA was the lowest.The order of D value and rD value was reversed.②The AUC of K value,D value,rK value and rD value for differential diagnosis of BT and MT was 0.722,0.691,0.687 and 0.608,respectively.Taking K value of 0.66,D value of 1.38×10-3mm2/s,rK value of-0.44,and rD value of 0.44 as the TVs,the SE was 66.95%,70.34%,58.47%and 46.61%,and the SP was 82.00%,64.00%,76.00%and 76.00%,respectively.The AUC of K value,D value,rK value and rD value for differential diagnosis of PA and WT was 0.879,0.893,0.775 and 0.796,respectively.Using the K value of 0.89,D value of 1.32×10-3mm2/s,rK value of-0.38,and rD value of 0 as the TVs,the SE was 100.00%,89.04%,88.52%and 86.30%,and the SP was 75.86%,79.31%,53.65%and 65.52%,respectively.The AUC of K value in the differential diagnosis of PA and BCA was 0.839.Using the K value of 0.66 as the TV,the SE was 61.64%,and the SP was 100.00%.The AUC of K value,D value,rK value and rD value for differential diagnosis of PA and MT was 0.886,0.786,0.759 and 0.676,respectively.Using the K value of 0.66,D value of 1.38×10-3mm2/s,rK value of-0.44,and rD value of 0.44 as the TVs,the SE was 87.67%,86.30%,69.86%and 54.79%,and the SP was 82.00%,64.00%,76.00%and 76.00%,respectively.The AUC of K value,D value and rD value for differential diagnosis of WT and BCA was 0.765,0.884 and 0.812,respectively.Using the K value of 0.88,D value of 1.08×10-3mm2/s,and rD value of 0.09 as the TVs,the SE was 81.82%,100.00%and 90.91%,and the SP was 75.86%,65.52%and 68.97%,respectively.The AUC of D value in the differential diagnosis of BCA and MT was 0.767,Using the D value of 1.39×10-3mm2/s as the TV,the SE was 81.82%,and the SP was 66.00%.5.T2 mapping analysis of parotid tumors:①The T2 value and rT2 value of MT were all significantly lower than those of BT(all P<0.001).PA had the highest T2 value and rT2 value,followed by BCA and MT,WT was the lowest.②The AUC of T2 value and rT2 value for differential diagnosis of BT and MT was 0.715 and 0.692,respectively.Using the T2 value of 108.22ms and rT2 value of 0.77 as the TVs,the SE was 56.78%and 57.63%,and the SP was 84.00%and 78.00%,respectively.The AUC of T2 value and rT2 value for differential diagnosis of PA and WT was 0.891 and 0.862,respectively.Using the T2 value of 93.21ms and rT2 value of 0.75 as the TVs,the SE was 90.41%and 78.08%,and the SP was 89.66%and 93.10%,respectively.The AUC of T2 value and rT2 value for differential diagnosis of PA and BCA was 0.686 and 0.664,respectively.Taking the T2 value of 145.91ms and the rT2 value of 1.56 as the TVs,the SE was 41.10%,and the SP was 100.00%.The AUC of T2 value and rT2 value for differential diagnosis of PA and MT was 0.836 and 0.799,respectively.Taking the T2 value of 102.73ms and the rT2 value of 0.77 as the TVs,the SP was 78.00%,and the SE was 80.82%and 76.71%,respectively.The AUC of T2 value in the differential diagnosis of WT and BCA was 0.865.Using the T2 value of 93.21ms as the TV,the SE was 81.82%,and the SP was 89.66%.6.T1 mapping analysis of parotid tumors:①The difference in ΔT1 between BT and MT was significant(P<0.05).PA had the highest Tlpre value,ΔT1,T1 change rate and rTlpre value,followed by BCA and MT,WT was the lowest.②The AUC ofΔT1 in the differential diagnosis of BT and MT was 0.618.Using the ΔT1 of 1459.06ms as the TV,the SE was 51.16%,and the SP was 80.95%.The AUC of Tlpre value,ΔT1,T1 change rate and rTlpre value for differential diagnosis of PA and WT was 0.900,0.945,0.925 and 0.835,respectively.Taking Tlpre value of 1768.41ms,ΔT1 of 931.02ms,T1 value change rate of 0.45,and rTlpre value of 1.25 as the TVs,the SE was 78.08%,85.96%,96.49%and 75.34%,and the SP was 39.10%93.33%,80.00%and 79.31%,respectively.The AUC of Tlpre value,ΔT1,T1 change rate and rTlpre value for differential diagnosis of PA and MT was 0.642,0.726,0.736 and 0.638,respectively.Using the T1pre value of 1943.62ms,ΔT1 of 1277.00ms,T1 value change rate of 0.71,and rTlpre value of 2.08 as the TVs,the SE was 72.60%,71.93%,50.88%and 52.05%,and the SP was 52.00%,71.43%,90.48%and 76.00%,respectively.The AUC of T1pre value,ΔT1,T1 change rate and rTlpre value for differential diagnosis of WT and BCA was 0.846,0.889,0.852 and 0.856,respectively.Using the Tlpre value of 2158.31ms,ΔT1 of 1249.79ms,T1 change rate of 0.55,and rTlpre value of 1.25 as the TVs,the SE was 72.73%,77.78%,88.89%and 81.82%,and the SP was 100.00%,100.00%,86.67%and 79.31%,respectively.The AUC of Tlpre value,ΔT1,T1 change rate and rTlpre value for differential diagnosis of WT and MT was 0.830,0.890,0.852 and 0.756,respectively.Taking the Tlpre value of 1544.67ms,ΔT1 of 578.54ms,T1 value change rate of 0.45,and rTlpre value of 0.69 as the TVs,the SE was 82.76%,80.00%,80.00%and 62.07%,and the SP was 74.00%,90.48%,90.48%and 82.00%,respectively.7.DCE-MRI analysis of parotid tumors:①There were significant differences in TIC types between the groups of parotid tumors(all P<0.05).The AUC of the TIC differentiate BT from MT,PA from WT,PA from BCA,and PA from MT was 0.667,0.756,0.800,and 0.729,respectively.The SE was 81.82%,71.19%,71.19%,and 71.19%,and the SP was 50.56%,100.00%,100.00%,and 81.82%,respectively.②The AUC of Ktrans value,Kep value,Ve value,rKtrans value and rKep value for differential diagnosis of PA and WT was 0.909,0.968,0.802,0.773 and 0.965,respectively.Using the Ktrans value of 0.20/min,Kep value of 0.68/min,Ve value of 0.22,rKtrans value of 1.41,and rKep value of-0.33 as the TVs,the SE was 83.05%,93.22%,83.05%,83.05%and 94.92%,and the SP was 87.50%,93.75%,81.25%,68.75%and 93.75%,respectively.The AUC of Kep value and rKep value for differential diagnosis of PA and BCA was 0.991 and 0.985,respectively.Using the Kep value of 0.85/min and rKep value of-0.33 as the TVs,the SE was 96.61%and 94.92%,and the SP was 100.00%,respectively.The AUC of Ktrans value,Kep value,Ve value and rKep value for differential diagnosis of PA and MT was 0.672,0.786,0.650 and 0.836,respectively.Using the value of 0.10/min,Kep value of 0.49/min,Ve value of 0.28,and rKep value of 4.04 as the TVs,the SE was 57.63%,72.88%,55.93%and 33.90%,and the SP was 77.27%,77.27%,70.45%and 81.82%,respectively.The AUC of Ktrans value,Kep value and rKep value for differential diagnosis of WT and MT were 0.814,0.872 and 0.832,respectively.Taking the Ktrans value of 0.34/min,Kep value of 1.92/min,and rKep value of 0.29 as the TVs,the SE was 75.00%,68.75%and 81.25%,and the SP was 84.09%,97.73%and 96.36%,respectively.8.Evaluation of the application value of the combined diagnostic model for parotid tumors:rADC,rK,rD,and rT2 were included in the combined diagnostic model,and the AUC for distinguishing BT from MT was 0.684,the SE was 84.00%,and the SP was 46.90%.rADC,rK,rD,rT2,T1 change rate,rT1pre,rKtrans,and rKep were included in the combined diagnostic model,and the AUC for distinguishing PA from WT was 0.995,the SE was 94.50%,and the SP was 100.00%.rKep was included in the combined diagnostic model,and the AUC for distinguishing PA from BCA was 0.960,the SE was 97.26%,and the SP was 81.82%.rADC,rK,rD,rT2,T1 change rate,rTlpre,and rKep were included in the combined diagnostic model,and the AUC for distinguishing PA from MT was 0.900,the SE was 80.00%,and the SP was 91.80%.rADC,rD,T1 change rate,and rTlpre were included in the combined diagnostic model,and the AUC for distinguishing WT from BCA was 0.975,the SE was 100.00%,and the SP was 93.10%.rADC,T1 change rate,rTlpre,and rKep were included in the combined diagnostic model,and the AUC for distinguishing WT from MT was 0.964,the SE was 86.00%,and the SP was 100.00%.rADC was included in the combined diagnostic model,and the AUC for distinguishing BCA from MT was 0.795,the SE was 48.00%,and the SP was 100.00%.Conclusions1.RESOLVE-DWI,RESOLVE-DKI,T2 mapping,T1 mapping and DCE-MRI can reveal great significance in the differential diagnosis of parotid gland tumors.Among them,DCE-MRI and T1 mapping are the preferred sequences in distinguishing Warthin tumor from malignant tumor.2.The combined diagnostic models can distinguish pleomorphic adenoma from Warthin tumor,pleomorphic adenoma from basal cell adenoma,pleomorphic adenoma from malignant tumor,Warthin tumor from basal cell adenoma,and Warthin tumor from malignant tumor.They can reveal high diagnostic efficiency and clinical application value. |
PDF Full Text Request