一种增强显现模式的癌症分类算法  

Cancer classification algorithm with improved emerging pattern

作　　者：卢新国[1] 李丹[1] 王海军[2] 

机构地区：[1]湖南大学计算机与通信学院,长沙410082 [2]河南科技大学理学院,河南洛阳471003

出　　处：《计算机工程与应用》2010年第26期233-237,共5页Computer Engineering and Applications

基　　金：国家自然科学基金No.60873184;湖南省自然科学基金No.07JJ5085~~

摘　　要：基于基因表达谱的癌症检测对癌症的治疗具有十分重要的意义。显现模式(Emerging Patterns,EPs)能够挖掘隐含的具有生物意义的基因表达模式,基于显现模式的癌症分类方法可以检测出癌症样本。针对提取显现模式时在小样本情况下将频率近似于概率的缺陷,引入贝叶斯估计以提高熵的可靠度,提取癌症表达中的增强显现模式(Improved Emerging Patterns,EPIs),并提出一种基于增强显现模式的基因分类算法(ACancer Classification Algorithm with Improved Emerging Patterns CCEPI)。最后在急性白血病数据集上进行实验,实验结果表明算法提高了癌症检测的正确率。Cancer detection with gene expression profiles is important for cancer treatment.Emerging Patterns（EPs） can discover the hidden gene expression patterns related to the cancer.And EPs based cancer classification algorithm can identify the cancer for the sample.For shortcoming in the estimation of probability with frequency when the size of samples is small in the EPs mining, Bayes estimation is introduced into the mining for increasing the reliability of entropy, then the Improved Emerging Patterns（EPIs） is extracted in cancer expressing,and a cancer classification algorithm with Improved Emerging Patterns（CCEPI） is proposed.The experiment is taken on the Leukemia dataset and the results show CCEPI can improve the correction in cancer detection.

关 键 词：显现模式 癌症分类 基因表达谱 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP391[自动化与计算机技术—控制科学与工程]