肺部孤立性阴影计算机辅助诊断专家系统

Expert System in the Aided Diagnosis of Pulmunary Solitary Nodule

模拟专家对肺部孤立性阴影诊断的思维过程以人工智能原理建立的辅助诊断专家系统可以提高肺癌,良性肿瘤及结核瘤诊断的正确率。病人信息分为临床指标及胸片上≤4cm直径的孤立性阴影信息。该系统采用规则编成分层结构的知识库,应用似然推理机制作出诊断;并采用PROLOG语言进行编程,具有解释功能和机器学习功能,用户界面友好,菜单式汉字提示。经415例临床验证总符合率达到94％,其中肺癌为95％,良性肿瘤92％,结核瘤96％。

In order to raise the accuracy of diagnosis in pulmonary solitary nodule no bigger than 4cm, an expert system, based on Artificial Interlligence theory was designed. The patient information includes clinical and roentgenographic parameters. The knowledge base of the system was constructed by a number of rules. The diagnostic process is realized by uncertainly inference. This system was programmed through micro-PROLOG language with functions of interpretation, inferences and machine learning, friendly interface and Chinese character menu prompting. Results show the mean accurate rate of the computer aided diagnosis is 94%, in which brochogenic carcinoma is 95%, benign tumour and tuberculoma. are 92% and 96% respectively.