摘要
目前国际上有关肺癌的呼吸检测诊断方法越来越受到关注,其具有简便、快速、无创伤、无标记以及非接触等显著特点。在前期大量肺癌呼出气体挥发性有机物(VOCs)标志物研究的基础上,采用基于MOS-SAW传感器联用的电子鼻技术,分析27例肺癌患者和27例健康人呼出气体样本,并对数据进行处理和识别,设计PCA、PLS、LDA以及ANN等多种诊断方法,比较不同算法的识别结果。实验结果表明,采用的人工神经网络复合模型对肺癌和健康人群的识别灵敏度和特异性分别达到92.59%和88.89%。所提出的复合识别方法对于电子鼻快速诊断肺癌患者是有效的。通过呼出气体中冷凝物标志物的检测和复合诊断算法,将进一步提高通过呼吸气体标志物诊断的新型电子鼻仪器在临床诊断中的广泛应用。
The method of breath detection for lung cancer has gained growing interests in recent years.It is simple,fast,non-invasive,no need of marks and non-contact.Based on the lung cancer biomarkers obtained from the volatile organic compounds in breath samples of lung cancer patients and controls,we analyzed 27 breath samples of lung cancer patients and 27 breath samples of controls using MOS-SAW sensor coupling electronic nose technique.After pre-processing and analysis of the data,multiple methods,including PCA,LDA,PLS and ANN,were employed to establish the diagnosis models.And the discrimination results of these models were compared.It was showed that the ANN method had the best performance with the sensitivity of 92.59% and the specificity of 88.89%.So this method can be considered as a candidate one for the diagnosis of lung cancer through breath detection.
出处
《中国生物医学工程学报》
CAS
CSCD
北大核心
2012年第1期110-116,共7页
Chinese Journal of Biomedical Engineering
基金
国家自然科学基金重点项目仪器专项(81027003)
教育部博士点基金(20100101110079)
浙江大学基本科研业务费专项资助