Auxiliary diagnosis method of myocardial infarction based on fusion of statistical features and entropy features

doi:10.11772/j.issn.1001-9081.2019071172

Journal of Computer Applications ›› 2020, Vol. 40 ›› Issue (2): 608-615.DOI: 10.11772/j.issn.1001-9081.2019071172

• Frontier & interdisciplinary applications • Previous Articles Next Articles

Auxiliary diagnosis method of myocardial infarction based on fusion of statistical features and entropy features

Zhizhong WANG¹, Longlong QIAN¹, Chuang HAN¹, Li SHI²()

^1.College of Electrical Engineering，Zhengzhou University，Zhengzhou Henan 450000，China
^2.Department of Automation，Tsinghua University，Beijing 100000，China

Received:2019-07-08 Revised:2019-08-17 Accepted:2019-08-27 Online:2019-09-11 Published:2020-02-10
Contact: Li SHI
About author:WANG Zhizhong， born in 1982， Ph. D.， associate professor. His research interests include biological signal detection and processing.
QIAN Longlong， born in 1993， M. S. candidate. His research interests include ECG signal analysis， intelligent diagnosis.
HAN Chuang， born in 1991， Ph. D. candidate. His research interests include ECG signal analysis， intelligent diagnosis.
Supported by:
the National Natural Science Foundation of China(61673353);the National Natural Science Foundation of China Young Scientists(61603344);the Henan Province Higher Education Key Research Project(15A120017)

基于统计特征和熵特征融合的心肌梗死辅助诊断方法

王治忠¹, 钱龙龙¹, 韩闯¹, 师丽²()

^1.郑州大学电气工程学院，郑州 450000
^2.清华大学自动化系，北京 100000

通讯作者: 师丽
作者简介:王治忠（1982—），男，山东蓬莱人，副教授，博士，主要研究方向：生物信号检测与处理
钱龙龙（1993—），男，河南周口人，硕士研究生，主要研究方向：心电信号分析、智能诊断
韩闯（1991-），男，河南驻马店人，博士研究生，主要研究方向：心电信号分析、智能诊断；
基金资助:
国家自然科学基金资助项目(61673353);国家自然科学基金青年科学基金资助项目(61603344);河南省高等教育重点研究项目(15A120017)

Abstract

Abstract:

Aiming at the problem of low clinical practicability and accuracy in the clinical diagnosis of myocardial infarction， an auxiliary diagnosis method of myocardial infarction based on 12-lead ElectroCardioGram （ECG） signal was proposed. Firstly， denoising and data enhancement were performed on the 12-lead ECG signals. Secondly， aiming at the ECG signals of each lead， the statistical features including standard deviation， kurtosis coefficient and skewness coefficient were extracted respectively to reflect the morphological characteristics of ECG signals， meanwhile the entropy features including Shannon entropy， sample entropy， fuzzy entropy， approximate entropy and permutation entropy were extracted to characterize the time and frequency spectrum complexity， the new mode generation probability， the regularity and the unpredictability of the ECG signal time series as well as detect the small changes of ECG signals. Thirdly， the statistical features and entropy features of ECG signals were fused. Finally， based on the random forest algorithm， the performance of algorithm was analyzed and verified in both intra-patient and inter-patient modes， and the cross-validation technology was used to avoid over-fitting. Experimental results show that， the accuracy and F1 value of the proposed method in the intra-patient modes are 99.98% and 99.99% respectively， the accuracy and F1 value of the proposed method in the inter-patient mode are 94.56% and 97.05% respectively； and compared with the detection method based on single-lead ECG， the detection of myocardial infarction with 12-lead ECG is more logical for doctors’ clinical diagnosis.

Key words: myocardial infarction, statistical feature, entropy feature, random forest algorithm, cross-validation, 12-lead ElectroCardioGram （ECG）

摘要：

针对心肌梗死临床诊断过程中临床实用性和准确率不高的问题，提出一种基于12导联心电图（ECG）的心肌梗死的辅助诊断方法。首先，对12导联ECG信号进行去噪和数据增强处理；其次，分别对各导联ECG信号提取包含标准差、峰度系数、偏度系数的统计特征，以此反映信号的形态特征；同时，提取包含香农熵、样本熵、模糊熵、近似熵和排列熵的熵特征，以此表征ECG信号时间序列的时间与频谱复杂性、新模式产生的概率、规律性和不可预测性以及检测ECG信号的微小变化；然后，融合ECG信号的统计特征和熵特征；最后，基于随机森林算法在病人内和病人间两种模式下对算法进行分析和验证，并通过交叉验证防止过拟合。实验结果表明，病人内模式下算法准确率和F1值分别为99.98%和99.99%，病人间模式下算法准确率和F1值分别为94.56%和97.05%；与基于单导联ECG的诊断方法相比，采用12导联ECG诊断心肌梗死更符合医生临床诊断逻辑。

关键词: 心肌梗死, 统计特征, 熵特征, 随机森林算法, 交叉验证, 12导联心电图

CLC Number:

TP391

Zhizhong WANG, Longlong QIAN, Chuang HAN, Li SHI. Auxiliary diagnosis method of myocardial infarction based on fusion of statistical features and entropy features[J]. Journal of Computer Applications, 2020, 40(2): 608-615.

王治忠, 钱龙龙, 韩闯, 师丽. 基于统计特征和熵特征融合的心肌梗死辅助诊断方法[J]. 《计算机应用》唯一官方网站, 2020, 40(2): 608-615.

Figures/Tables 13

References 26

1	ACHARYA U R， KANNATHAL N， LEE M H， et al. Study of heart rate variability signals at sitting and lying postures［J］. Journal of Bodywork and Movement Therapies， 2005， 9（2）： 134-141. 10.1016/j.jbmt.2004.04.001
2	钟清华，易子川. 基于Ⅰ导联ECG的心肌梗死特征提取［J］. 生物医学工程学杂志， 2013， 30（2）： 260-264.
	ZHONG Q H， YI Z C. Feature extraction of myocardial infarction based on lead I ECG［J］. Journal of Biomedical Engineering， 2013， 30（2）：260-264.
3	常战国，蒲宝明，李相泽，等. 机器学习实现心肌梗死的自动检测［J］. 计算机系统应用， 2019， 28（4）： 218-224.
	CHANG Z G， PU B M， LI X Z， et al. Machine learning achieves automatic detection of myocardial infarction［J］. Computer Systems and Applications， 2019， 28（4）： 218-224.
4	武会江，黄启俊，常胜，等. 基于多项式参数拟合和支持向量机的心肌梗死识别算法［J］. 中国医学物理学杂志， 2017， 34（7）： 736-738. 10.3969/j.issn.1005-202X.2017.07.017
	WU H J， HUANG Q J， CHANG S， et al. Myocardial infarction recognition algorithm based on polynomial fitting and support vector machine［J］. Chinese Journal of Medical Physics， 2017， 34（7）：736-738. 10.3969/j.issn.1005-202X.2017.07.017
5	PADHY S， DANDAPAT S. Third-order tensor based analysis of multilead ECG for classification of myocardial infarction［J］. Biomedical Signal Processing and Control， 2017， 31：71-78. 10.1016/j.bspc.2016.07.007
6	张清丽，苏士美，王猛. 基于改进小波变换的QRS特征提取算法研究［J］. 郑州大学学报（理学版）， 2017， 49（4）： 100-103. 10.13705/j.issn.1671-6841.2016340
	ZHANG Q L， SU S M， WANG M. QRS feature extraction algorithm based on wavelet transform［J］. Journal of Zhengzhou University （Natural Science Edition）， 2017， 49（4）：100-103. 10.13705/j.issn.1671-6841.2016340
7	REMYA R S， INDIRADEVI K P， BABU K K A. Classification of myocardial infarction using multi resolution wavelet analysis of ECG［J］. Procedia Technology， 2016， 24：949-956. 10.1016/j.protcy.2016.05.195
8	SHARMA L N， TRIPATHY R K， DANDAPAT S. Multiscale energy and eigenspace approach to detection and localization of myocardial infarction［J］. IEEE Transactions on Biomedical Engineering， 2015， 62（7）： 1827-1837. 10.1109/TBME.2015.2405134
9	BANERJEE S， MITRA M. Application of cross wavelet transform for ECG pattern analysis and classification［J］. IEEE Transactions on Instrumentation and Measurement， 2014， 63（2）： 326-333. 10.1109/tim.2013.2279001
10	ACHARYA U R， FUJITA H， ADAM M， et al. Automated characterization and classification of coronary artery disease and myocardial infarction by decomposition of ECG signals： a comparative study［J］. Information Sciences， 2017， 377：17-29. 10.1016/j.ins.2016.10.013
11	SHARMA M， TAN R S， ACHARYA U R. A novel automated diagnostic system for classification of myocardial infarction ECG signals using an optimal biorthogonal filter bank［J］. Computer in Biology and Medicine， 2018， 102：341-356. 10.1016/j.compbiomed.2018.07.005
12	LEE J， JEONG K， YOON J， et al. A simple real-time QRS detection algorithm［C］// Proceedings of 18th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. Piscataway： IEEE， 1996：1396-1398. 10.1109/iembs.1996.647473
13	DOHARE A K， KUMAR V， KUMAR R. Detection of myocardial infarction in 12 lead ECG using support vector machine［J］. Applied Soft Computing， 2018， 64：138-147. 10.1016/j.asoc.2017.12.001
14	ARIF M， MALAGORE I A， AFSAR F A. Detection and localization of myocardial infarction using K-nearest neighbor classifier［J］. Journal of Medical Systems， 2012， 36（1）： 279-289. 10.1007/s10916-010-9474-3
15	BHASKAR N A. Performance analysis of support vector machine and neural networks in detection of myocardial infarction［J］. Procedia Computer Science， 2015， 46：20-30. 10.1016/j.procs.2015.01.043
16	CHANG P C， LIN J J， HSIEH J C， et al. Myocardial infarction classification with multi-lead ECG using hidden Markov models and Gaussian mixture models［J］. Applied Soft Computing， 2012， 12（10）： 3165-3175. 10.1016/j.asoc.2012.06.004
17	ACHARYA U R， FUJITA H， OH S L， et al. Application of deep convolutional neural network for automated detection of myocardial infarction using ECG signals［J］. Information Sciences， 2017， 415/416：190-198. 10.1016/j.ins.2017.06.027
18	KUMAR M， PACHORI R B， ACHARYA U R. Automated diagnosis of myocardial infarction ECG signals using sample entropy in flexible analytic wavelet transform framework［J］. Entropy， 2017， 19（9）：No.488. 10.3390/e19090488
19	SADHUKHAN D， PAL S， MITRA M. Automated identification of myocardial infarction using harmonic phase distribution pattern of ECG data［J］. IEEE Transactions on Instrumentation and Measurement， 2018， 67（10）： 2303-2313. 10.1109/tim.2018.2816458
20	凌振宝，王大任，蔡靖. 基于差分阈值法和BP神经网络算法的心肌梗死诊断［J］. 中国老年学杂志， 2009， 39（2）：1540-1543.
	LIN Z B， WANG D R， CAI J. Diagnosis of myocardial infarction based on differential threshold method and BP neural network algorithm［J］. Chinese Journal of Gerontology， 2009， 39（2）：1540-1543.
21	HAN C， SHI L. Automated interpretable detection of myocardial infarction fusing energy entropy and morphological features［J］. Computer Methods and Programs in Biomedicine， 2019， 175：9-23. 10.1016/j.cmpb.2019.03.012
22	GOLDBERGER A L， AMARAL L A， GLASS L， et al. PhysioBank， PhysioToolkit， and PhysioNet： components of a new research resource for complex physiologic signals［J］. Circulation， 2000， 101（23）：215-220. 10.1161/01.cir.101.23.e215
23	SUN L， LU Y， YANG K， et al. ECG analysis using multiple instance learning for myocardial infarction detection［J］. IEEE Transactions on Biomedical Engineering， 2012， 59（12）： 3348-3356. 10.1109/TBME.2012.2213597
24	ACHARYA U R， FUJITA H， SUDARSHAN V K， et al. Automated detection and localization of myocardial infarction using electrocardiogram： a comparative study of different leads［J］. Knowledge-Based Systems， 2016， 99：146-156. 10.1016/j.knosys.2016.01.040
25	STRODTHOFF N， STRODTHOFF C. Detecting and interpreting myocardial infarction using fully convolutional neural networks［J］. Physiological Measurement， 2019， 40（1）： 41-51. 10.1088/1361-6579/aaf34d
26	LIU W， HUANG Q， CHANG S， et al. Multiple-feature-branch convolutional neural network for myocardial infarction diagnosis using electrocardiogram［J］. Biomedical Signal Processing and Control， 2018， 45： 22-32. 10.1016/j.bspc.2018.05.013

MI				HC
病人号		数据增强前记录数	数据增强结果	病人号	数据增强前记录数	数据增强结果
1		3	374	104	1	114
2		1	534	105	1	143
3		1	303	116	1	114
4		2	833	117	2	250
…		…	…	…	…	…
290		1	84	267	1	106
291		1	107	276	1	147
292		2	119	277	1	95
293		2	117	279	4	364
294		1	100	284	3	431
总数	142	367	50 818	52	77	9 912

MI				HC
病人号		数据增强前记录数	数据增强结果	病人号	数据增强前记录数	数据增强结果
1		3	374	104	1	114
2		1	534	105	1	143
3		1	303	116	1	114
4		2	833	117	2	250
…		…	…	…	…	…
290		1	84	267	1	106
291		1	107	276	1	147
292		2	119	277	1	95
293		2	117	279	4	364
294		1	100	284	3	431
总数	142	367	50 818	52	77	9 912

信号类型	样本总数	训练集样本数	测试集样本数
MI	50 020	45 018	5 002
HC	9 990	8 991	999

信号类型	样本总数	训练集样本数	测试集样本数
MI	50 020	45 018	5 002
HC	9 990	8 991	999

真实情况	预测结果
真实情况	MI	HC
MI	TP（真正例）	FN（假反例）
HC	FP（假正例）	TN（真反例）

Auxiliary diagnosis method of myocardial infarction based on fusion of statistical features and entropy features

基于统计特征和熵特征融合的心肌梗死辅助诊断方法

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 13

References 26

Related Articles 11

Recommended Articles

Metrics

分类器	折数	真实值	病人内预测结果		病人间预测结果		acc（病人内/间）/%	sen（病人内/间）/%	spe（病人内/间）/%	F1（病人内/间）/%
分类器	折数	真实值	0	1	0	1	acc（病人内/间）/%	sen（病人内/间）/%	spe（病人内/间）/%	F1（病人内/间）/%
BPNN	1	0	5 002	0	5 464	174	100.00	100.00	100.00	100.00
	1	1	0	999	270	259	92.80	96.91	48.96	96.09
	2	0	5 002	0	5 881	2	100.00	100.00	100.00	100.00
	2	1	0	999	99	420	98.42	99.96	80.92	99.14
	3	0	5 001	1	7 497	110	99.98	99.98	100.00	99.99
	3	1	0	999	216	282	95.97	98.55	56.62	97.87
	4	0	5 002	0	5 924	149	100.00	100.00	100.00	100.00
	4	1	0	999	159	378	95.34	97.54	70.39	97.46
	5	0	5 002	0	3 180	225	100.00	100.00	100.00	100.00
	5	1	0	999	5	534	94.16	93.39	99.07	96.50
	6	0	5 001	1	7 107	586	99.98	99.98	100.00	99.99
	6	1	0	999	297	277	89.31	92.38	48.25	94.15
	7	0	4 992	10	6 270	1154	99.83	99.80	100.00	99.89
	7	1	0	999	125	454	84.01	84.45	78.41	90.74
	8	0	5 002	0	2 623	62	99.98	100.00	99.90	99.99
	8	1	1	998	122	387	94.23	97.69	76.03	96.61
	9	0	5 002	0	1 339	157	99.98	100.00	99.90	99.99
	9	1	1	998	140	372	85.20	89.50	72.65	90.01
	10	0	5 002	0	1 754	151	99.96	100.00	99.80	99.98
	10	1	2	997	278	210	82.07	92.07	43.03	89.10
RF	1	0	5 002	0	5 606	32	99.95	100.00	99.69	99.97
	1	1	3	996	280	249	94.94	99.43	47.06	97.29
	2	0	5 002	0	5 883	0	100.00	100.00	100.00	100.00
	2	1	0	999	4	515	99.93	100.00	99.22	99.96
	3	0	5 002	0	7 600	7	99.95	100.00	99.69	99.97
	3	1	3	996	371	127	95.33	99.90	25.50	97.57
	4	0	5 002	0	6 025	48	99.98	100.00	99.89	99.99
	4	1	1	998	203	334	96.20	99.20	62.19	97.95
	5	0	5 002	0	3 362	43	99.95	100.00	99.69	99.97
	5	1	3	996	253	286	92.49	98.73	53.06	95.78
	6	0	5 002	0	7 522	171	99.96	100.00	99.79	99.98
	6	1	2	997	299	275	94.31	97.77	47.90	96.97
	7	0	5 002	0	7 175	249	99.98	100.00	99.89	99.99
	7	1	1	998	177	402	94.67	96.64	69.43	97.11
	8	0	5 002	0	2 685	0	99.98	100.00	99.89	99.99
	8	1	1	998	236	273	92.61	100.00	53.63	95.79
	9	0	5 002	0	1 446	50	99.98	100.00	99.89	99.99
	9	1	1	998	274	238	83.86	96.65	46.48	89.92
	10	0	5 002	0	1 884	21	100.00	100.00	100.00	100.00
	10	1	0	999	277	211	87.47	98.89	43.23	92.67
KNN	1	0	4 971	31	5 055	583	99.31	99.38	98.99	99.58
	1	1	10	989	249	280	86.50	89.65	52.93	92.39
	2	0	4 978	24	5 182	701	99.53	99.52	99.59	99.71
	2	1	4	995	122	397	87.14	88.08	76.49	92.64
	3	0	4 977	25	6 921	686	99.40	99.50	98.89	99.36
	3	1	11	988	397	101	86.63	90.98	20.28	92.74
	4	0	4 974	28	4 999	1074	99.38	99.44	99.09	99.62
	4	1	9	990	268	269	79.69	82.31	50.09	88.16
	5	0	4 981	21	3 329	76	99.53	99.58	99.29	99.71
	5	1	7	992	318	221	90.01	97.76	41.00	94.41
	6	0	4 967	35	6 745	948	99.31	99.30	99.39	99.58
	6	1	6	993	294	280	84.97	87.67	48.78	91.56
	7	0	4 985	17	5 969	1455	99.56	99.66	99.09	99.73
	7	1	9	990	12	567	81.66	80.40	97.92	89.05
	8	0	4 982	20	2 033	652	99.55	99.60	99.29	99.72
	8	1	7	992	402	107	67.00	75.71	21.02	79.41
	9	0	4 989	13	1 496	0	99.63	99.74	99.09	99.78
	9	1	9	990	158	354	92.13	100.00	69.14	94.98
	10	0	4 977	25	1 720	185	99.45	99.50	99.19	99.66
	10	1	8	991	336	152	78.22	90.28	31.14	86.84

来源文献	核心方法	导联数	交叉验证	性能（病人内/间）
文献［14］	QRS波检测，时域特征，KNN	12导联	否	sen=99.97%/无； spe=99.90%/无
文献［23］	ST段检测，多示例学习，SVM	12导联	否	sen=92.60%/无； spe=82.40%/无
文献［24］	用47个特征进行心肌梗死诊断，KNN	12导联	否	acc=98.80%/无； sen=99.45%/无； spe=96.27%/无
文献［17］	CNN	II导联	否	acc=93.53%/无； sen=93.71%/无； spe=92.83%/无
文献［18］	FAWT和Sent特征，LS-SVM	I导联	否	acc=99.31%/无；
文献［25］	SVM	8导联	否	sen=93.30%/无； spe=89.70%/无
文献［26］	MFB-CNN	12导联	否	acc=99.95%/98.79%
文献［19］	相位特征，阈值分类器	3导联	否	acc=95.60%/无； sen=96.50%/无； spe=92.70%/无
本文方法	统计特征和熵值特征，RF、BPNN、KNN	12导联	是	acc=99.98%/99.93%； sen=100%/100%； spe=99.85%/99.22%； F1=99.99%/99.96%

[1]	Jingcheng XU, Xuebin CHEN, Yanling DONG, Jia YANG. DDoS attack detection by random forest fused with feature selection [J]. Journal of Computer Applications, 2023, 43(11): 3497-3503.
[2]	YU Dongchang, ZHAO Wenfang, NIE Kai, ZHANG Ge. Visibility forecast model based on LightGBM algorithm [J]. Journal of Computer Applications, 2021, 41(4): 1035-1041.
[3]	ZHU Yuna, ZHANG Yutao, YAN Shaoge, FAN Yudan, CHEN Hantuo. Protocol identification approach based on semi-supervised subspace clustering [J]. Journal of Computer Applications, 2021, 41(10): 2900-2904.
[4]	HU Cong, QU Jinjin, XU Chuanpei, ZHU Aijun. Garment image recognition based on adaptive pooling neural network [J]. Journal of Computer Applications, 2018, 38(8): 2211-2217.
[5]	ZHOU Ying, FANG Yong, HUANG Cheng, LIU Liang. Detection of SQL injection behaviors for PHP applications [J]. Journal of Computer Applications, 2018, 38(1): 201-206.
[6]	ZHOU Shuangshuang, XU Jin'an, CHEN Yufeng, ZHANG Yujie. New words detection method for microblog text based on integrating of rules and statistics [J]. Journal of Computer Applications, 2017, 37(4): 1044-1050.
[7]	XIU Chao, CAO Lin, WANG Dongfeng, ZHANG Fan. Automatic lane division method based on echo signal of microwave radar [J]. Journal of Computer Applications, 2017, 37(10): 3017-3023.
[8]	TIAN Shuowei, YANG Yuexiang, HE Jie, WANG Xiaolei, JIANG Zhixiong. Real-time detection system for stealthy P2P hosts based on statistical features [J]. Journal of Computer Applications, 2015, 35(7): 1892-1896.
[9]	WANG Shu, LYU Xueqiang, ZHANG Kai, LI Zhuo. Image classification approach based on statistical features of speed up robust feature set [J]. Journal of Computer Applications, 2015, 35(1): 224-230.
[10]	WANG Yi-gang, ZHU Xiao-dong, GAN Mao-zhi. Pattern recognition method for critical modules of software based on neural networks [J]. Journal of Computer Applications, 2005, 25(06): 1336-1338.
[11]	DU Hui-qian, MEI Wen-bo, SU Guang-chuan. Image interpolation using wavelet transform based on statistical features [J]. Journal of Computer Applications, 2005, 25(02): 332-334.