基于极限学习机的左束支传导阻滞辅助诊断研究

发布时间：2018-03-28 00:00

本文选题：极限学习机　切入点：左束支传导阻滞　出处：《中国生物医学工程学报》2017年03期

【摘要】：左束支传导阻滞(LBBB)作为临床常见的一种心律失常,是左心室收缩功能减低、患者死亡率增加的标志;利用机器学习算法对其进行辅助诊断,将对LBBB早发现、早治疗起到积极的推动作用。然而,由于目前常用的支持向量机(SVM)等传统的机器学习算法容易产生局部最优解,准确度有待提高,因此提出一种基于极限学习机(ELM)的LBBB辅助诊断算法。首先,利用小波进行心电信号预处理,包括基线漂移、肌电噪声及工频干扰的去除;接着,确定QRS波群与T波位置;然后,根据临床上LBBB患者比正常人的QRS波群持续时间延长等特点,建立融合时域、形态与能量3类特征的特征模型;最后,利用该模型提取的特征集合,提出基于ELM的LBBB辅助诊断算法。此外,在MIT_BIH数据库中的5 000份ECG数据上进行实验验证,结果表明所提出的预处理与波形提取算法能有效去除噪声并提取QRS-T特征波;在LBBB的判别上,相比SVM算法、ELM算法的训练时间缩短了88.5%;同时,在准确率、灵敏度、特异度、LBBB检出率和正常人检出率的指标上,分别提升2.4%、5.4%、1.2%、3.6%和2%。因此,基于ELM的LBBB辅助诊断算法具有明显优势。
[Abstract]:Left bundle branch block (LBBB), as a common arrhythmia in clinic, is a marker of decreased left ventricular systolic function and increased mortality in patients. Machine learning algorithm is used to assist diagnosis and early detection of LBBB. However, traditional machine learning algorithms such as support vector machine (SVM) are easy to produce local optimal solutions, and the accuracy needs to be improved. Therefore, a LBBB aided diagnosis algorithm based on ELM) is proposed. Firstly, the ECG signal is preprocessed by wavelet, including baseline drift, myoelectric noise and power frequency interference removal. Then, the position of QRS wave group and T wave is determined. According to the longer duration of QRS wave group in LBBB patients compared with normal people, the feature model of fusion time domain, shape and energy is established. Finally, the feature set extracted by this model is used. A LBBB aided diagnosis algorithm based on ELM is proposed. In addition, the experimental results on 5 000 ECG data in MIT_BIH database show that the proposed preprocessing and waveform extraction algorithm can effectively remove noise and extract QRS-T characteristic waves. Compared with SVM algorithm, the training time of LBBB algorithm is reduced by 88.5.The accuracy, sensitivity, specificity of LBBB detection rate and the detection rate of normal people are increased by 2.4%, 5.42% and 3.6% and 2%, respectively, when compared with that of SVM algorithm. Therefore, the accuracy, sensitivity, specificity of LBBB and the detection rate of normal people are increased by 2.4% and 3.2%, respectively. The LBBB aided diagnosis algorithm based on ELM has obvious advantages.
【作者单位】：东北大学中荷生物医学与信息工程学院;东北大学计算机科学与工程学院;大连医科大学附属第二医院;
【基金】：国家自然科学基金青年项目(61402089) 辽宁省自然科学基金面上项目(2015020553) 中国博士后科学基金项目(2016M591447)
【分类号】：R541.7;TP181

【相似文献】