项目名称: 肺癌血清数据的模式识别解析与诊断及其特征标记物的探寻
项目编号: No.21305043
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 数理科学和化学
项目作者: 李艳坤
作者单位: 华北电力大学(保定)
项目金额: 25万元
中文摘要: 人体血清含有的丰富信息能为癌瘤的诊断和治疗提供重要线索。但目前基于血清的临床癌瘤的早期筛查方法,存在灵敏度低、准确率低、特异性差的问题。项目拟针对这些问题,展开对肺癌血清的化学信息(计量)学方法的研究。对健康人和肺癌患者血清的拉曼光谱和多项成分指标数据进行解析,实现对(亚型)肺癌较准确地早期诊断。目前,此项相关工作有待展开系统的研究。项目拟采用非负矩阵分解等对非相关线性判别分析(ULDA)方法进行改进,寻找最佳分类子空间及特征变量,并结合改进的稳健多模型共识算法,提高识别的准确率和稳定性;研究适于血清数据的癌瘤特征信息提取方法(联合预处理方法、变量筛选新判据);并与主成分分析等方法的识别结果进行综合评价。以期建立稳健可靠、实用的基于模式识别技术的新型拉曼光谱(亚型)肺癌诊断方法和(亚型)肺癌血清多指标决策模型,探寻潜在的有临床价值或应用前景的(亚型)肺癌的特异性标记物及提供相应理论依据。
中文关键词: 肺癌;模式识别;拉曼光谱;多指标决策模型;特征标记物
英文摘要: Human serum contains abundant information, which can provide important clues to the diagnosis and treatment of cancer. But the early screening method of cancer based on serum has the problems of low sensitivity, low accuracy and poor specificity. To solve the above problems, the project will develop the research of chemoinformatics (chemometrics) methods on lung cancer serum. Raman spectra and multiple components data on serums of healthy people and lung cancer patients will be analyzed, and comparatively accurate early clinical diagnosis of lung cancer or lung cancer subtype are expected. At present, the related research work is expected to carry out systematically. In the project, ULDA algorithm will be improved by non-negative matrix factorization etc. for searching the best classification subspace and feature variables, and furthermore, improved robust consensus modeling algorithm will be combined to improve the accuracy and stability of model recognition; Feature information extraction methods (combined pretreatment methods and new variable selection criterions) for cancer serum data will be proposed; And the results of principal component analysis method etc. will be compared for comprehensive evaluations of pattern recognition methods. The reliable and practical new Raman spectra model and serum multiple
英文关键词: Lung cancer;Pattern recogition;Raman spectra;multiple attribute decision model;Tumor marker