论文部分内容阅读
目的：使肝外胆管梗阻病变（ＥＢＯＤ）的ＣＴ鉴别诊断智能化。方法：对１１７例ＥＢＯＤ的ＣＴ影像资料和临床数据进行计算机分析，建立本研究资料ＳＡＳ数据库，并用逐步选择法对１６项因素筛选。然后对诊断有影响的１０项数值变量和分类变量进行判别分析。按每一分类变量对诊断的影响程度分别赋于权值。数值变量直接输入判别过程，最后得到两个线性判别方程。结果：用判别方程进行回代分析，良性梗阻病变正确判别率８５．６％，恶性８９．９％。将这一结果编制成软件，并前瞻性地对另５４例病例进行判别分析，良性病变正确判别率８１．８％，恶性７８．１％。结论：计算机专家辅助系统增加了ＥＢＯＤ鉴别诊断的准确性，实现了信息输入，权值转化，模拟诊断和结果输出功能。
Objective: To make the diagnosis of extrahepatic bile duct obstruction (EBOD) CT differential diagnosis intelligent. Methods: The CT images and clinical data of 117 cases of EBOD were analyzed by computer. SAS data of this study was established and 16 factors were screened by stepwise selection. Then the diagnosis of the impact of the 10 numerical variables and classification variables for discriminant analysis. The impact of each categorical variable on diagnosis is assigned to weights, respectively. Numerical variables are directly input to the discriminant process, and finally two linear discriminant equations are obtained. Results: Using discriminant equation for back analysis, the correct discrimination rate of benign obstructive lesions was 85.6% and malignant 89.9%. The results compiled into software, and prospectively discriminate the other 54 cases, the correct identification rate of benign lesions 81.8%, malignant 78.1%. Conclusion: Computer expert assistant system increases the accuracy of EBOD differential diagnosis, and realizes the functions of information input, weight conversion, simulation diagnosis and result output.