卵巢恶性肿瘤彩色多普勒超声诊断的Logistic模型分析

A logistic regression model for predicting ovarian malignant tumor with color Doppler ultrasound

目的筛选卵巢恶性肿瘤的独立恶性征象,评价彩色多普勒超声对其的诊断价值。方法回顾性分析经手术后病理证实的128个卵巢肿块(良性63个、恶性65个)的超声表现,通过Logistic回归分析筛选独立诊断指标,建立Logistic模型。结果最后进入Logistic模型的3个特征分别为肿块壁厚薄及形态(P<0.001)、肿块血供(P<0.001)和内部回声(P=0.049)。结论肿块壁厚而不规则或有突起、血供丰富、内部回声不均是卵巢恶性肿瘤的独立特征指标。彩色多普勒超声Logistic回归模型有助于卵巢恶性肿瘤的鉴别诊断。

Objective To evaluate the color Doppler ultrasonic features of malignant ovarian tumors, and to assess the accuracy of ultrasonic diagnosis of ovarian tumors. Methods The ultrasonic features of 65 malignant and 63 benign ovarian tumors were retrospectively reviewed. A logistic model for predicting malignant ovarian tumors was obtained. Results Three ultrasonic features were finally entered the logistic regression model. They were tumor walls （P〈0.001）, blood supply （P〈0.001） and internal echoes （P=0.049）. Conclusion Thickening tumor walls, abundant blood supply and unequal internal echoes are main characteristics of malignant ovarian tumors. The logistic regression model can efficiently differentiate malignant from benign ovarian tumors.