Regression Analysis Of Multmodal Ultrasonic Features Of Pleomorphic Adenomas And Warthin Tumors

Objective:Logistic regression analysis was used to evaluate the diagnostic value of gray-scale ultrasound combined with color doppler flow imaging(CDFI)and contrast-enhanced ultrasound(CEUS)in pleomorphic adenomas and Warthin tumors.The differential factors between pleomorphic adenoma of salivary glands and Warthin tumor were screened out and a multimodal ultrasound diagnosis model was established.Method:From August 2015 to July 2018,75 patients with benign tumor of salivary glands were enrolled in the Affiliated Hospital of Qingdao University.The results of multimodal ultrasound were analyzed retrospectively and compared with those of surgical pathology.?~2 tests were used to compare the ultrasonographic characteristics of pleomorphic adenoma of salivary glands with Warthin tumor.The parameters with statistical significance were taken as independent variable x,pathological results as dependent variable Y,and three diagnostic models were established according to the possible combination of multimodal ultrasound:model I includes the characteristics of gray-scale ultrasound,model II includes the characteristics of gray-scale ultrasound and color doppler flow imaging,model III includes the characteristics of gray-scale ultrasound,color doppler flow imaging and contrast-enhanced ultrasound.Binary logistic regression was performed.The receiver operating characteristic curve(ROC)was established,and the area under the curve(AUC)was calculated and the logistic regression analysis was performed.Result:(1)This study included 75 patients with a total of 82 lesions,including 45pleomorphic adenomas and 37 Warthin tumors.There were 7 factors were significant differences between pleomorphic adenomas and Warthin tumors in the two-dimensional gray-scale ultrasound features(morphology,non-echoic area,strong echo),CDFI grade and CEUS features(enhanced intensity,enhanced distribution characteristics and curve type)(?~2=13.718?12.551,3.915,20.045,31.303,8.829,6.480,P<0.05).(2)The above seven variables with statistical significance were assigned respectively,morphology:X1,non-echoic area:X2,strong echo:X3,CDFI grade:X4,enhancement intensity:X5,enhanced distribution characteristics:X6,curve type:X7,pathological result:Y.Three logistic regression models were statistically significant(?~2=0.78,0.78,0.78,P<0.05).Logistic regression equation:model I:logistic(P):19.766-1.766X1+1.970X2-19.704X3;model II:logistic(P):12.806-1.241X1+2.128X2-18.548X3+1.938X4;model III:logistic(P):20.646-0.915X1+1.761X2-18.051X3+0.620X4+1.108X5-1.818X6-1.295X7.Through logistic regression analysis,it was found that the effective factors to distinguish pleomorphic adenoma and Warthin tumor were the non-echoic area and enhanced intensity(?~2=1.47,5.78,P<0.05).(3)The sensitivity of model?,?,?to the detected lesions was 83.8%,94.6%,97.3%,the specificity was 64.4%,60.0%,84.4%,and the accuracy was 70.1%,70.2%,86.1%.The area under ROC of three logistic regression models was 0.793,0.871,0.935(P<0.01).Conclusion:(1)There are two important factors in the differential diagnosis of pleomorphic adenoma and Warthin tumor in salivary glands:the non-echo area and the enhanced intensity.(2)The ultrasound multimodal diagnosis model is better than the single modality diagnostic model in terms of diagnostic performance,and ultrasound multimodal diagnosis model can improve the accuracy of diagnosis and discrimination of pleomorphic adenoms and Warthin tumors in salivary glands.