|
|
|
| Application of the binary Logistic regression model to analyze the results of benign and malignant superficial lymph nodes diagnosed by ultrasonography |
| HAN Zhi-yu, LI Xin, SHAO Qiu-jie |
| Department of Ultrasound, General Hospital of PLA, Beijing 100853, China |
|
|
|
|
Abstract Objective: To apply the binary Logistic regression model to analyze the ultrasonogrphic differential diagnosis of benign and malignant superficial lymph nodes. Methods: The ultrasonography of superficial lymph nodes of 171 patients proved by pathology and follow-up were analyzed, and we selected the indexes of differential diagnosis through the model of binary Logistic regression analysis to see if these were statistical significance. Results: All indexes of ultrasonography were analyzed by the multivariable binary Logistic regression model through the forward stepwise. Some of those indexes were selected out and induced in the equation, which included: having the normal lymph hilum or not, the recode by the ratio of longitudinal over short diameter, having the area of acoustic transmission or not, the signal of the blood flow rich or scarce, the signal types of the blood flow, and the covariates of age and sex． Moreover, all those indexes were included of the indexes which had significant statistical differences using χ2 test. Meanwhile, the results shown that the malignant probability of a lymph node would change with altering of the signal types of blood flow. Conclusion: The multivariate analysis model of binary Logistic regression can describe and analyze the process of differential diagnosis of malignant and benign lymph nodes by ultrasonography, and can select out the valuable indexes of differential diagnosis.
|
|
Received: 27 July 2006
|
|
|
|
|
|
|
2007, Vol. 18 
