The Study On The Classification And Staging Of Endometrial Cancer Based On Conventional And Functional MRI Texture Analysis

Objective:1.To investigate the diagnostic value of apparent diffusion coefficient?ADC?value and quantitative parameters of dynamic contrast-enhanced MRI?DCE-MRI??K^transrans value,Kep value and Ve value?in the classification and staging of endometrial cancer.2.To investigate the diagnostic value of texture parameters based on FS-T2WI,DWI and dynamic contrast enhanced magnetic resonance imaging(?DCE-MRI?in the classification and staging of endometrial cancer.Methods:Experiment 1:a total of 60 patients with endometrial cancer who were confirmed by surgery and pathology in the first affiliated hospital of wannan medical college from October 2017 to August 2019 were collected.All patients underwent routine and functional?DWI and DCE?MRI one week prior to surgery.Using post-processing workstation and GE OK software to obtain quantitative parameters;All layers of lesions on MRI images were manually sketched and fused into 3d ROI,and the texture features were extracted by AK software of GE company.According to the immunohistochemical results,there were 38 cases of estrogen-dependent?type I?and 22 cases of non-estrogen-dependent?type II?.Referring to the 2009 FIGO staging method,the results of surgical and pathological staging of all patients were as follows:40 cases of early endometrial cancer?24 cases of stage Ia,16 cases of stage Ib?,and 20 cases of advanced endometrial cancer?stage II-IV?.By using two independent sample t-test,the Mann-Whitney U test to compare statistical difference of different classification,staging,and the depth of muscular invasion of endometrial cancer ADC values,DCE MRI quantitative parameter values(K^transrans value,V_e and K_epp value)and texture parameters,using multiple logistic regression analysis model,ROC curve drawing,evaluate effectiveness of ADC values,DCE-MRI quantitative diagnostic parameters,texture parameters and its model in different classification,staging,and depth of muscular invasion of endometrial carcinoma.Results:DWI and Quantitative parameters of DCE:1.K^transrans value of type II endometrial cancer was higher than that of type I,and the difference was statistically significant?P=0.006?.ADC value,Kep value and Ve value were not significantly different between type I and type II?P>0.05?.The ROC area under the ROC curve for the diagnosis of endometrial cancer typing by K^transrans value was0.754?P=0.001?.K^trans=0.43/min was taken as the threshold,and the sensitivity and specificity of the diagnosis of endometrial cancer typing were 93.75%and 52.17%,respectively.2.ADC value,K^transrans value,K_epp value and V_e value of endometrial cancer in stage Ia and Ib showed no significant difference?P>0.05?.3.Ve value of advanced endometrial cancer was higher than that of the early group,and the difference was statistically significant?P=0.011?.ADC value,K^transrans value and K_epp value were not significantly different between the early group and the advanced group?P>0.05?.The area under the ROC curve of Ve value in diagnosing endometrial cancer stage was 0.713?P=0.007?.Ve=0.55/min was taken as the threshold,and the sensitivity and specificity for endometrial cancer stage were 63.75%and 83.33%,respectively.Texture analysis and its model:AK software extracted 1044 texture features from FS-T2WI,DWI and DCE-MRI sequences,totaling 3,132 texture features.1.After feature sifting,there were statistically significant differences in 13 parameters between type I and type II endometrial cancer.The AUC value of texture parameter was0.330?0.748,in which DCE?Percentile5 had the highest AUC value?0.743?.The AUC value of optimal model for multivariate logistic regression analysis in diagnosing endometrial cancer classification was 0.810.2.After feature sifting,there were statistically significant differences in 8 parameters between stage Ia and Ib endometrial carcinoma.The AUC value for differentiating stage Ia and Ib endometrial cancer was 0.270?0.736,among which DCE?MinIntensity was the most effective?AUC value was 0.736?,and The AUC value of optimal model for multivariate logistic regression analysis in differentiating stage Ia and Ib endometrial cancer was 0.780.3.After feature sifting,there were statistically significant differences in 11 parameters between early and advanced endometrial carcinoma.The AUC value of early and progressive endometrial carcinoma was 0.233?0.758,among which T2?HaralickCorrelation?AllDirection?offset5?SD was the most effective?AUC value was 0.758?,and the AUC of the best model of multivariate logistic regression analysis in differentiating early and advanced endometrial carcinoma was 0.903.Conclusion:1.DCE-MRI quantitative parameters K^transrans value,texture parameters and their models all have certain value in the classification diagnosis of endometrial cancer,and the value of texture parameter model is higher.2.All the quantitative parameters of DWI and DCE-MRI could not identify stage Ia and Ib endometrial carcinoma.Texture parameters and their models are valuable in the diagnosis of stage Ia and Ib endometrial carcinoma.3.DCE-MRI quantitative parameter Ve value is helpful for the diagnosis of early and advanced endometrial cancer;Texture parameters and their models are valuable in the differentiation of early and advanced endometrial carcinoma.4.Texture parameters and their models were superior to DWI and DCE-MRI quantitative parameters in differentiating endometrial cancer classification,stage Ia and Ib,and early and advanced stages.