摘要
目的探讨数据挖掘中决策树模型在MRI诊断腮腺区肿块中的应用。方法连续纳入2013年12月至2016年10月接受MRI检查的腮腺区占位的患者共32例,其中良性肿块24例,恶性肿块8例。分别取2项临床指标和9项MRI检查征象信息用于建立决策树。结果成功建立了能够判断腮腺区肿块良恶性的决策树模型,其中含有4条诊断规则,最低相对错误代价为0.0625,对诊断具有决策意义的最重要的前3项决策指标为肿块的边界、与邻近血管的关系以及与周围组织分界情况。结论对于MRI诊断腮腺区肿块,决策树可以对其良恶性进行正确地判断,有助于对医师进行规则指导,提高诊断水平,为未来计算机辅助诊断疾病提供有效的参考信息及途径。
Objective To explore the application of decision tree model in MRI diagnosis of mass in the parotid gland area. Methods Two clinical indicators and nine MRI findings were used to establish the decision tree. From December 2013 to October 2016,32 patients with masses in the parotid gland area who received MRI examination were included( benign masses,24 cases; malignant masses,8 cases). Results A decision tree model was established to determine the benign and malignant mass in the parotid gland area,which contained four diagnostic rules. The lowest relative error cost was 0. 0625. The most important top decision index for diagnosis was the boundary of the masses,the relationship between the masses and the adjacent blood vessels,the demarcation of the masses with the surrounding tissue. Conclusion For the diagnosis of parotid gland mass by MRI,the decision tree can prove reasonable judgments about the benign and malignant masses in the parotid gland area,and it will help to guide the doctors and improve the diagnosis level,thus,provide effective reference information and path for the future computer-aided diagnosis.
出处
《临床放射学杂志》
CSCD
北大核心
2018年第2期210-213,共4页
Journal of Clinical Radiology
基金
广西医疗卫生适宜技术开发与推广应用项目(编号:S201628)
关键词
腮腺区肿块
磁共振成像
决策树
Mass in the parotid gland area MRI Decision tree