A Logistic Regression Analysis on Ultrasonic Differential Diagnosis of Small Breast Nodules

Objective To investigate the features by analyzing the sonograms of small breast nodules(≤15 mm).The points for differential diagnosis of breast cancer through Logistic regression analysis were selected and risk rank was disposed to improve the accurate rate of differential diagnosis of small breast cancer by Ultrasound.Methods 166 cases of small breast nodules(≤15 mm)confirmed by pathological examination were collected and retrospectively analyzed,a Logistic model for predicting breast malignancy on the basis of ultrasonographic features was obtained and the risk degrees were ranked through OR values.Results The Logistic regression analysis demonstrated that 9 ultrasonic features were enrolled in the Logistic equation,Indistinct and discontinue of cooper ligament,super fascia and lobule of breast(9.182),angular margin(7.675),microcalcification(7.471),shape taller than wide(4.776),spicular sign(3.982),absence of lateral shadow(2.277),posterior attenuation(1.861),uneven echogenicity(1.640),irregular shape(1.097).Conclusion The Logistic model of sonographic features is helpful to differentiate the benign small nodules from the malignant ones,and has significant value in differential diagnosis of small breast cancer.