MR-T_(2)WI纹理分析评价盆底功能障碍性疾病耻骨内脏肌损伤的价值

MR-T_(2)WI texture analysis in quantitative evaluation of pelvic floor dysfunction value of pubovisceral muscle injury

目的探讨MR-T_(2)WI纹理分析评价盆底功能障碍性疾病耻骨内脏肌损伤的价值。方法选取94例已育女性T_(2)WI图像,其中54例盆底功能障碍性疾病患者为观察组,40例盆底功能正常者为对照组。使用ITK-SNAP软件在T_(2)WI图像获得耻骨内脏肌全域三维感兴趣区,使用A.K.分析软件提取纹理特征。使用Mann Whitney U检验、单因素Logistic回归和最小冗余最大相关(mRMR)特征选择方法进行特征筛选和降维,同时筛选临床风险因素,共同构建多因素Logistic回归模型,绘制受试者工作特征曲线(ROC),并计算曲线下面积(AUC)值。最后利用交叉验证评价模型效能。结果共提取了1219个纹理特征,去冗余后剩余7个纹理特征,其中wavelet_HHL_firstorder_Minimum AUC值最高0.87,一个临床特征Age,AUC值为0.84,其余特征的AUC值均大于0.75。将这些特征纳入多因素逻辑回归分析,建立联合预测模型,其模型敏感度和特异度分别为87%及98%,AUC值为0.96。结论MR-T_(2)WI纹理分析的部分定量参数可评价盆底功能障碍性疾病耻骨内脏肌损伤,有助于临床诊疗。

Objective To evaluate the value of MR-T_(2)WI texture analysis in pelvic floor dysfunction disease of pubovisceral muscle injury. Methods A retrospective analysis was conducted on 94 fertile women's T_2WI images, in whom 54 patients with pelvic floor dysfunction were in the observation group, and 40 women with normal pelvic floor function were in the control group. We used ITK-SNAP software to obtain the full three-dimensional region of interest of the pubovisceral muscles on the T_2WI image, and used A.K. analysis software to extract texture features. We used Mann Whitney U test, single factor Logistic regression and maximum relevance minimum redundancy(mRMR) for feature selection and dimensionality reduction, while screened clinical risk factors, constructed a multivariate Logistic regression model, and plotted the subjects the receiver operating characteristic(ROC) and calculated the area under curve(AUC) value. Finally, cross-validation was used to verify the effectiveness of the model. Results A total of 1 219 texture features were extracted, and 7 texture features remained after de-redundancy. Among them, the wavelet_HHL_firstorder_Minimum AUC value was the highest 0.87, one clinical texture feature, age, the AUC value was 0.84, and the AUC values of the other features were all greater than 0.75. These characteristics were incorporated into multiple logistic regression analysis to establish a predictive model. The sensitivity and specificity of the model were 87% and 98%, respectively, and the AUC value was 0.96. Conclusion MR-T_(2)WI texture analysis can quantitatively evaluate pelvic floor dysfunction of pubovisceral muscle injury, which is helpful for clinical diagnosis and treatment.