【作者】 黎琪；

【导师】 田逢春；

【作者基本信息】 重庆大学 ， 信号与信息处理， 2008， 硕士

【摘要】 乳癌是人类最常见的一种恶性肿瘤,也是女性主要恶性肿瘤之一,发病率逐年上升,乳癌的早期诊断与治疗,有利于提高五年存活率。传统的乳癌诊断方法,一般采用X光扫描,活检等方法实现,检测程序复杂,周期较长而不利于乳癌的早期诊断与治疗,本课题开展基于电子鼻的乳癌诊断研究,方法简便,易于实现,对乳癌的早诊断早治疗极为有利。本文首先从病理学角度对乳腺癌医用电子鼻系统的可行性进行了论证,阐述了医用电子鼻的发展现状,介绍了医用电子鼻信号处理中的信号预处理方法以及模式识别方法,探讨了基于电子鼻的无创诊断方法在呼吸气体诊断中的应用优势。根据有关乳癌呼吸诊断研究资料,结合乳癌发病的病理学依据,论文选定了壬烷(nonane)、异丙醇(iso-Propyl alcohol)、苯乙酮(1-phenyl-Ethanone)、正庚醛(Heptanal)、肉豆蔻酸异丙酯(isopropyl myristate)等挥发性产物作为基于电子鼻的乳癌呼吸诊断的主要挥发性标志物。然后围绕乳癌呼吸诊断标志物,进行了气体传感器的调查研究,最终选定了六支金属氧化物气体传感器、一支电化学气体传感器和温度、湿度、压力传感器构建了乳癌诊断电子鼻系统。由于乳癌呼吸诊断标志物浓度较低(ppb级),要求乳癌诊断电子鼻的传感器具有很低的LoD(最低检测限);而在目前的技术条件下,普通气体传感器都达不到这个要求,上述气体传感器阵列无法直接检测临床病人呼吸气体,论文进一步研究了气体浓缩方法,在参照有关国际标准的基础上,着重考虑了精密的温度控制和流量控制的实现,构建了一个可重复性较好的固体吸附、热解吸气体浓缩系统。最后,利用构建的乳癌电子鼻系统,对乳癌呼吸诊断的部分挥发性标志物进行了测试,并利用BP网络和RBF网络进行了定性识别,BP网络和RBF网络对三种气体27个样本的识别率均为100%,分类效果良好,验证了该系统对乳癌呼吸诊断标志物的敏感能力。更多还原

【Abstract】 Breast cancer is one of the most common human malignancies, and also is one of the major women malignancies. The incidence of breast cancer increased year by year, and the early diagnosis and treatment of breast cancer would significantly improve the five-year survival rate. Traditional diagnosis methods of breast cancer generally adopt X-ray scanning and live organization examination, the diagnosis procedure is relatively complex and long, which is adverse for early breast cancer diagnosis and treatment. In this project, we carried out the research for the diagnosis of the breast cancer based on electronic nose. The method is simple and easy to implement, and the research is helpful for the early diagnosis and treatment of breast cancer.From the point of view of breast cancer pathology, this paper first verifies the feasibility of medical electronic nose system. Then we introduce the state of the art of medical electronic nose systems, and medical electronic nose signal preprocessing methods and pattern recognition methods. After that, we explore the the application advantange of electronic nose based lossless diagnosis methods for the respiratory gas diagnosisAccording to the previous research of the respiratory diagnosis of breast cancer, combined with the pathology principle of breast cancer, we choose nonane, iso-Propyl alcohol, 1-phenyl-Ethanone, heptanal and isopropyl myristate as the volatile symbol of electronic nose based breast cancer respiratory diagnosis. We focused on the volatile symbol of breast cancer respiratory diagnosis to conduct researches and surveys. At last, we selected six metal oxide gas sensors, an electrochemical sensor, and temperature, humidity and pressure sensors to build a breast cancer diagnosis electronic nose system.Because of the low density of the volatile symbol of breast cancer respiratory diagnosis (ppt level), it is required that the sensors of electronic nose of breast cancer diagnosis can provide low LoD(Limit of Detection). However, under current technique condition, normal gas sensors can not achieve this requirement. And the above gas sensor arrays are unable to diagnose the respiratory of patient’ respiratory gas. This paper describes the study of the gas concentration, and constructs a solid adsorption, thermal desorption gas concentration system, which is based on the international stardand and the consideration of the implementation of precise temperature control and throughput control. Finally, we conducted the test of the volatile symbol of breast cancer respiratory diagnosis based on the constructed breast cancer electronic nose system. In addition, we utilized the BP artificial neural network and RBF network to perform qualitative reorganization.Experiments had good effect, which verified the sensitive capacity of the system for the volatile symbol of breast cancer respiratory diagnosis.更多还原