DNA算法优化BP网络用于心律失常的识别.doc

DNA算法优化BP网络用于心律失常的识别 巾I组纵I究与临床康短掰74蕾392010—09—24出版 JournalofClinicalRehabilitativeTissueEngineeringResearchSeptember24,2010Vol14No.39 DNA算法优化BP网络用于心律失常的识别料☆ 师黎,赵云,郭豹 ArrhythmiarecognitionusingaBPneuralnetwork-basedDNAalgorithm ShiLi.ZhaoYun,GuoBao Abstract BACKGROUND:Arrhythmiaiscommonlydiagnosedbydoctors.andthediagnosisdependsondoctorexperienceand responsibility.Arrhythmiarecognitionisgreatimportantforrescuingandearlytreatmentofcardiopathysufferers 0BJECTlVE:ToInvestigateefficientmethodofauto—recognizinganddiagnosisofarrhythmia METHODS:Thewholemorphologyofarrhythmicheartbeatwasextractedfromelectrocardiograph(ECG).anddiscretecosine transformandjnversediscretecosinetransformwereusedtocompressthedata.ABPneuralnetworkwasdesignedfor arrhythmicheartbeatrecognition,andtheinitialweightsandthresholdsofnetworkwereoptimizedbyDNAalgorithm.Finally.the MIT,BIHECGdatabasewasusedtotesttheDNA—BPneuraInetwork RESULTSANDCONCLUS10N:Forthefivetypesheartbeat,includingnormal,Ieftbundlebranchblock,rightbundlebranch blockventricularescapeheartbeat,Pacedheartbeat,theexperimentresultsdemonstrateefficientbyusingDNA—BPneural network.withanaveragerecognitionrateof99% ShiL,ZhaoY,GuoB.ArrhythmiarecognitionusingaBPneuralnetwork—basedDNAalgorithm.ZhongguoZuzhiGongcheng YanjiuyuLinchuangKangfu.2010:14(39):7353—7357.【http://www.crter.ca】 摘要 背景:目前对心律失常的诊断大多是由医生人工完成,费时费力,诊断结果依赖于医生的个人业务水平和责任心.心律失 常的自动识别对于心脏病患者的救护和早期治疗具有非常重要的意义. 目的:实现临床心律失常的自动识别和诊断. 方法:首先从心电图中动态提取完整心律失常心拍形态,并采用离散余弦变换和反变换压缩数据:然后设计用于心律失常 识别的BP神经网络,并用DNA算法优化该BP网络;最后用MIT/BIH心电数据库中心电图数据对DNA-BP网络进行检验. 结果与结论:对于5种心拍类型,包括正常,左束支阻滞,右束支阻滞,心室跳脱心搏及Paced心搏,利用DNA.BP网 络进行分类,实验达到了很好的识别效果,平均识别正确率达到99%. 关键词:心电图;心律失常;自动识别:BP网络:DNA算法 doi:10.3969~issn.1673-8225.2010.39.032 schooIofElectricaI Engineering, Zhengzhou University, Zhengzhou450001 HenanProvince China ShiLi☆.Doctor. DoctoraIsupervisor, Professor,SchooIof ElectricaI Engineering, Zhengzhou University, Zhengzhou450001 HenanProvince. China shill@zzueducn CorresDondenceto ZhaoYun,Master. SchooIofElectrica