心音信号分析与自动诊断系统研究

分类号密级UDC学号092021222桂林电子科技大学硕士学位论文题目______________心音信号分析与自动诊断系统研究____（英文）StudyonAnalysisandAutomaticDiagnosisSystemofHeartSounds研究生学号:_______092021222______________研究生姓名:_________王晓燕_________指导教师姓名、职务:_________曾庆宁___教授_________申请学科门类:_________工学硕士_____________学科、专业:__________信号与信息处理_________提交论文日期:__________2012年4月____________论文答辩日期:__________2012年6月____________年月日独创性（或创新性）声明本人声明所呈交的论文是我个人在导师指导下进行的研究工作及取得的研究成果。尽我所知，除了文中特别加以标注和致谢中所罗列的内容以外，论文中不包含其他人已经发表或撰写过的研究成果；也不包含为获得桂林电子科技大学或其它教育机构的学位或证书而使用过的材料。与我一同工作的同志对本研究所做的任何贡献均已在论文中做了明确的说明并表示了谢意。申请学位论文与资料若有不实之处，本人承担一切相关责任。本人签名：日期：关于论文使用授权的说明本人完全了解桂林电子科技大学有关保留和使用学位论文的规定，即：研究生在校攻读学位期间论文工作的知识产权单位属桂林电子科技大学。本人保证毕业离校后，发表论文或使用论文工作成果时署名单位仍然为桂林电子科技大学。学校有权保留送交论文的复印件，允许查阅和借阅论文；学校可以公布论文的全部或部分内容，可以允许采用影印、缩印或其它复制手段保存论文。（保密的论文在解密后遵守此规定）本学位论文属于保密在____年解密后适用本授权书。本人签名：日期：导师签名：日期：摘要摘要心血管疾病出现临床症状前，心脏跳动会呈现异常情况，产生的心音信号中会出现心音分裂、心脏杂音和附加音，这些畸变和杂音成分中含有重要的生理病理信息。心脏听诊能够获得这些信息，但这项技能很难掌握。因此采集心音信号并以心音图的形式保存下来，再利用数字信号处理和模式识别技术进行心音分析具有重要意义。本文首先实现了基于FPGA的心音信号采集，该过程包括前级放大电路、带通滤波电路、后级放大电路、A/D转换电路和FPGA控制串口通信电路，采集到的心音信噪比高。然后对采集到的心音信号进行分析研究。文中先阐述了传统阈值消噪的原理以及存在的缺点，提出一种参数可调小波阈值函数消噪法，仿真实验表明该方法消噪效果明显优于传统阈值消噪法。接着采用归一化平均香农能（NASE）和简单度算法进行心音包络提取，提出了基于心音杂音边界线和简单度的分段算法，实验结果表明该算法分段准确率较高。昀后以LPCC、MFCC、PCA_MLLT作为输入特征参数，HMM作为识别模型进行心音分类研究，重点研究了HMM训练中，融合K均值聚类的Baum-Welch多观察序列参数重估模型训练算法，并利用自己建立的心音库来测试心音分类系统性能，实验结果表明该系统心音分类正确率较高。昀后开发了一款基于Matlab的数字心音分析诊断软件。该软件能够实现心音信号的小波阈值去噪、心音信号分析以及自动诊断功能，为下一步的科学研究、医学教学、临床应用奠定了基础。关键词：心音采集；小波阈值消噪；心音分析；心音分类；Matlab-I-AbstractAbstractThebeatingoftheheartpresentstheabnormalconditionbeforecardiovasculardiseasemanifestsclinicalsymptoms.Theproducedheartsoundmaybesplitedorcontainedheartmurmursandadditionalnoise.Thesemurmursincludeimportantphysiologyandpathologyinformation.Thecardiacauscultationcanacquirethisinformation.Butitisaverydifficultskillforpeopletomaster.Thus,ithasgreatsignificancetocollectheartsoundsignalandsaveitinthephonocardiogramformforlaterheartsoundanalysisusingdigitalsignalprocessingandpatternrecognitiontechnology.ThispaperfirstlyintroducesadesignofaheartsoundacquisitionsystembasedonFPGA.Itincludesfrontamplifiercircuit,band-passfiltercircuit,rearamplifiercircuit,A/DconvertercircuitandserialcommunicationcircuitcontrolledbyFPGA.ThecollectedheartsoundwithithashigherSNR.Secondly,studyingonthecollectedsignal,thispaperelaboratesthetraditionalthresholdde-noisingtheoryandtheexistingshortcomings,andthentheadjustableparameterswaveletthresholdfunctionisputforwardforde-noisingthecollectedheartsound.Theresultsshowthattheeffectsaremuchbetterthantraditionalthresholdde-noisingmethods.AndthenNormalizedAverageShannonEnergy(NASE)andSimplicityareusedtodrawheartsoundenvelope.Inaddition,anewsegmentationalgorithmisproposedwhichemploysheartmurmurborderandsimplicity.Experimentalresultsindicatethatthisalgorithmhashigheraccuracysegmentation.Moreimportantly,bytakingLPCC,MFCC,PCA_MLLTasinputcharacteristicparametersandHMMasrecognitionmodelforheartsoundsclassification,thispaperfocusesonthefusionofK-meansclusteringBaum-WelchparametersestimationalgorithmfortrainingtheHMMmodel.Aheartsoundlibrarycreatedbymyselfisusedtotestthesystemperformance.Theresultsshowthatthedevelopedsysteminthispaperhasahighaccuracyforheartsoundsclassification.Finally,digitalheartsoundanalysisanddiagnosissoftwarearedevelopedbasedonMatlab.Thissoftwarecanrealizethewaveletthresholdde-noisingofheartsounds,heartsoundsanalysisandautomaticdiagnosis,itiseasytooperateandcanbeappliedforpreliminaryanalysisandclassification.Besides,thissoftwaremaylayafoundationforfurtherscientificresearch,medicaleducationandclinicalapplication.Keywords:Heartsoundscollection;Waveletthresholdde-noising;Heartsoundssegmentation;Heartsoundsclassification;Matlab-II-目录目录摘要······································································································································IAbstract·······························································································································II第一章绪论·······················································································································1§1.1本课题的研究意义·································································································1§1.2国内外研究现状·····································································································3§1.3论文工作安排·········································································································4第二章基于FPGA的心音信号采集················································································6§2.1心音信号采集总体结构·························································································6§2.2心音传感器的研究·································································································6§2.2.1心音传感器的工作原理···················································································6§2.2.2基于驻极体话筒心音传感器的制作·······························································7§2.3心音信号预处理电路设计·····················································································9§2.3