利用关联规则挖掘恶性孤立性肺结节影像诊断规则

Using Association Rule Extracting Imaging Diagnosis Rules of Malignant Solitary Pulmonary Nodules

目的:探索数据挖掘关联规则算法对恶性SPN临床和影像诊断规则挖掘的价值,为计算机辅助诊断提供基础数据,为SPN影像诊断探索新的方法和思路。材料和方法:收集经肺穿刺或手术后病理学证实为肺癌的资料完整的恶性SPN 1 60例,对SPN临床及影像学表现属性进行标准化制定,设计出统一的规范化属性评价表,采取数据挖掘关联规则改进的Apriori算法对160例恶性SPN临床和CT表现属性进行了属性分层和属性合并的方式进行挖掘。结果:恶性SPN临床属性所挖掘出的诊断规则总共810条,恶性SPN影像一级属性所挖掘出的诊断规则总共780条,恶性SPN CT影像的一级、二级及其以上属性所挖掘出的诊断规则总共978条,恶性SPN临床和影像一级、二级及其以上属性所挖掘出的诊断规则总共822条。结论:关联规则算法可以用于恶性SPN诊断规则的数据挖掘的研究,其挖掘出来的诊断规则与临床公认的诊断规则是一致的,具有比较高的可信度。

Objective： To explore the value of association rules algorithm to the clinical and imaging diagnosis rule OT malignant solitary pulmonary nodules（SPN）ito supply the based data for computer-aided diagnosis of SPN and study the new methods and ideas for SPN. Materials and Methods： 160 cases of malignant SPN were confirmed by pathology, the uniform standardized evaluation form for the clinicat and imaging features of SPN was designed. 160 cases malignant SPN properties data of the clinical and CT manifestations were mined by the improving association rules apriori algorithm in the way of attribute layers and combined attributes. Results： The clinical attributes of malignant SPN extract 810 diagnostic rules, the primary CT imaging attributes of malignant SPN extract 780 diagnostic rules, the primary, the secondary and above CT imaging attributes of malignant SPN extract 978 diagnostic rules, the clinical and all imaging attributes of malignant SPN extract 822 diagnostic rules. Conclusion： The association rules can be used for the research of malignant SPN diagnosis rules, the diagnosis rules by data mining is consistent with the clinical diagnosis accepted rule, these diagnosis rules have relatively high credibility.