基于核稀疏編碼的陣發(fā)性房顫檢測

劉明; 孟憲輝; 熊鵬; 劉秀玲

doi:10.11999/JEIT190582

基于核稀疏編碼的陣發(fā)性房顫檢測

doi: 10.11999/JEIT190582 cstr: 32379.14.JEIT190582

1.
河北大學(xué)電子信息工程學(xué)院保定 071002
2.
河北省數(shù)字醫(yī)療工程重點(diǎn)實(shí)驗(yàn)室保定 071002

基金項(xiàng)目: 國家自然科學(xué)基金(61673158)，河北省自然科學(xué)基金(F2018201070)，河北省研究生創(chuàng)新資助項(xiàng)目(CXZZSS2019006)，河北省青年拔尖人才項(xiàng)目(BJ2019044)

詳細(xì)信息

作者簡介:
劉明：男，1972年生，博士，副教授，研究方向?yàn)槟Ｊ阶R別和心電信號處理

孟憲輝：女，1994年生，碩士生，研究方向?yàn)樾碾娦盘柼幚?/p>
熊鵬：女，1986年生，博士，講師，研究方向?yàn)槟Ｊ阶R別和生物信號處理

劉秀玲：女，1977年生，博士，教授，研究方向?yàn)樯镝t(yī)學(xué)成像和信號處理

通訊作者:
劉秀玲　liuxiuling121@hotmail.com

中圖分類號: TP399
計(jì)量
- 文章訪問數(shù): 2188
- HTML全文瀏覽量: 1157
- PDF下載量: 50
- 被引次數(shù): 0
出版歷程
- 收稿日期: 2019-08-01
- 修回日期: 2020-03-04
- 網(wǎng)絡(luò)出版日期: 2020-03-27
- 刊出日期: 2020-07-23

Detection of Paroxysmal Atrial Fibrillation Based on Kernel Sparse Coding

1.
College of Electronic and Information Engineering, Hebei University, Baoding 071002, China
2.
Key Laboratory of Digital Medical Engineering of Hebei Province, Baoding 071002, China

Funds: The National Natural Science Fundation of China (61673158), The Natural Science Foundation of Hebei Province (F2018201070), The Graduate Innovation Funding Project of Hebei Province (CXZZSS2019006), The Hebei Young Talent Project (BJ2019044)

摘要

摘要:
陣發(fā)性房顫(PAF)是一種具有偶發(fā)性的心律失常，其較高的漏檢率導(dǎo)致心臟相關(guān)疾病的增加。該文提出了一種基于核稀疏編碼的自動檢測方法，可以僅根據(jù)較短RR間期數(shù)據(jù)識別PAF發(fā)作。該方法采用特殊幾何結(jié)構(gòu)來分析數(shù)據(jù)高維特性，通過計(jì)算協(xié)方差矩陣作為特征描述子，找到蘊(yùn)含在數(shù)據(jù)中的黎曼流形結(jié)構(gòu)；然后基于Log-Euclid框架，利用核方法將流形空間映射到高維可再生核希爾伯特空間，以獲取更準(zhǔn)確的稀疏表示來快速識別PAF。經(jīng)麻省理工學(xué)院-貝斯以色列醫(yī)院房顫數(shù)據(jù)庫驗(yàn)證，獲得98.71%的敏感性、98.43%的特異度和98.57%的總準(zhǔn)確率。因此，該研究對檢測短暫發(fā)作的PAF有實(shí)質(zhì)性的改善，在臨床監(jiān)測和治療方面顯示出良好的潛力。
- 陣發(fā)性房顫 /
- 協(xié)方差描述子 /
- 黎曼流形 /
- 核稀疏編碼
Abstract:
Paroxysmal Atrial Fibrillation (PAF) is a kind of accidental arrhythmia, and its high missed detection rate leads to the increase of heart-related diseases. An automatic detection method is proposed based on kernel sparse coding, which can identify PAF attacks based only on short RR interval data. A special geometric structure is presented to analyze the high-dimensional characteristics of the data, and the covariance matrix is calculated as a feature descriptor to find the Riemannian manifold structure contained in the data; Based on the Log-Euclidean framework, a manifold method is used to map the manifold space to a high-dimensional renewable kernel Hilbert space to obtain a more accurate sparse representation to identify quickly PAF. After verification by the Massa-chusetts Institute of Technology-Beth Israel Hospital atrial fibrillation database, the sensitivity is 98.71%, the specificity is 98.43%, and the total accuracy rate is 98.57%. Therefore, this study has a substantial improvement in the detection of transient PAF and shows good potential for clinical monitoring and treatment.
- Paroxysmal Atrial Fibrillation(PAF) /
- Covariance descriptor /
- Riemann manifold /
- Kernel sparse coding

HTML全文

圖 1 PAF檢測流程圖

下載: 全尺寸圖片幻燈片

圖 2 PAF患者ECG記錄的RR間期時(shí)間序列

下載: 全尺寸圖片幻燈片

圖 3 PAF參數(shù)變化的所需計(jì)算時(shí)間

下載: 全尺寸圖片幻燈片

表 1 參數(shù)變化的檢測性能(%)

字典原子數(shù)(N)	重復(fù)交叉驗(yàn)證	分割滑動窗口(n)
		16			32			64
		Se	Sp	Acc	Se	Sp	Acc	Se	Sp	Acc
40	數(shù)據(jù)集1	97.99	96.63	97.32	98.44	97.95	98.19	98.86	97.75	98.30
	數(shù)據(jù)集2	97.95	97.42	97.68	98.74	98.15	98.44	98.67	98.43	98.55
	數(shù)據(jù)集3	98.00	97.99	97.99	98.65	98.51	98.51	98.97	98.30	98.64
	數(shù)據(jù)集4	97.38	98.44	97.91	98.50	98.67	98.59	98.78	98.57	98.67
	數(shù)據(jù)集5	98.36	98.34	98.35	98.49	98.55	98.52	98.89	98.57	98.73
	平均	97.94	97.76	97.85	98.56	98.37	98.45	98.83	98.32	98.58
60	數(shù)據(jù)集1	98.15	96.91	97.53	98.38	98.31	98.34	98.97	96.04	97.51
	數(shù)據(jù)集2	98.26	97.32	97.79	98.06	98.12	98.09	98.46	94.26	96.36
	數(shù)據(jù)集3	98.32	97.10	97.71	98.19	98.52	98.36	98.97	98.40	98.68
	數(shù)據(jù)集4	97.76	98.41	98.09	98.78	98.52	98.65	98.91	98.64	98.78
	數(shù)據(jù)集5	98.03	98.67	98.35	98.57	98.60	98.58	98.86	98.53	98.69
	平均	98.10	97.68	97.89	98.39	98.41	98.40	98.83	97.17	98.00
80	數(shù)據(jù)集1	98.15	97.26	97.70	98.52	98.24	98.38	98.97	98.17	98.57
	數(shù)據(jù)集2	97.99	97.43	97.71	98.81	98.27	98.54	98.97	98.35	98.66
	數(shù)據(jù)集3	97.98	97.96	97.97	98.86	98.31	98.58	99.00	98.48	98.74
	數(shù)據(jù)集4	97.39	98.24	97.81	98.73	98.66	98.69	98.93	98.44	98.69
	數(shù)據(jù)集5	97.60	98.62	98.11	98.65	98.66	98.65	98.88	98.67	98.78
	平均	97.82	97.90	97.86	98.71	98.43	98.57	98.95	98.42	98.69
100	數(shù)據(jù)集1	98.29	97.29	97.79	98.77	98.23	98.50	99.00	97.01	98.01
	數(shù)據(jù)集2	98.13	97.72	97.92	98.81	98.09	98.45	98.94	98.56	98.75
	數(shù)據(jù)集3	97.70	97.72	97.71	97.52	98.51	98.01	98.94	98.80	98.87
	數(shù)據(jù)集4	97.90	98.38	98.14	98.60	98.72	98.66	98.94	98.80	98.87
	數(shù)據(jù)集5	98.35	98.47	98.41	98.70	98.68	98.69	98.97	98.63	98.80
	平均	98.07	97.92	97.95	98.48	98.45	98.46	98.96	98.36	98.66
120	數(shù)據(jù)集1	96.64	94.03	95.33	97.87	95.86	96.86	98.89	97.10	97.99
	數(shù)據(jù)集2	98.11	93.22	95.66	98.81	97.74	98.28	97.73	97.84	97.79
	數(shù)據(jù)集3	97.59	97.49	97.04	98.79	98.54	98.66	99.00	97.36	98.18
	數(shù)據(jù)集4	98.24	98.10	98.17	98.50	98.59	98.54	98.94	98.46	98.70
	數(shù)據(jù)集5	98.23	98.44	98.34	98.34	98.68	98.51	98.80	98.54	98.67
	平均	97.76	96.26	96.91	98.46	97.88	98.17	98.67	97.86	98.27

下載: 導(dǎo)出CSV

表 2 不同算法分類效果對比(%)

文獻(xiàn)	年份	RR間期長度	Se	Sp	Acc
Lian等人^[20]	2011	128	95.89	95.40	–
Huang等人^[18]	2011	128	96.10	98.10	–
Petr?nas等人^[7]	2015	60	97.10	98.30	–
Zhou等人^[19]	2015	128	97.37	98.44	97.99
Cui等人^[8]	2017	150	97.04	97.96	97.78
Andersen等人^[10]	2018	31	98.98	96.95	97.80
本文方法	2019	33	98.71	98.43	98.57

下載: 導(dǎo)出CSV

參考文獻(xiàn)(20)

HAQQANI H M, CHAN K H, GREGORY A T, et al. Atrial fibrillation: State of the art in 2017-shifting paradigms in pathogenesis, diagnosis, treatment and prevention[J]. Heart, Lung and Circulation, 2017, 26(9): 867–869. doi: 10.1016/s1443-9506(17)31276-3

DE SISTI A, LECLERCQ J F, HALIMI F, et al. Evaluation of time course and predicting factors of progression of paroxysmal or persistent atrial fibrillation to permanent atrial fibrillation[J]. Pacing and Clinical Electrophysiology, 2014, 37(3): 345–355. doi: 10.1111/pace.12264

ZHOU Xiaolin, DING Hongxia, UNG B, et al. Automatic online detection of atrial fibrillation based on symbolic dynamics and Shannon entropy[J]. BioMedical Engineering OnLine, 2014, 13(1): 18. doi: 10.1186/1475-925X-13-18

SEPULVEDA-SUESCUN J P, MURILLO-ESCOBAR J, URDA-BENITEZ R D, et al. Atrial Fibrillation Detection Through Heart Rate Variability Using a Machine Learning Approach and Poincare Plot Features[M]. TORRES I, BUSTAMANTE J, and SIERRA D A. VII Latin American Congress on Biomedical Engineering CLAIB 2016, Bucaramanga, Santander, Colombia. Singapore: Springer Nature Singapore Pte Ltd, 2016: 565–568.

ANDERSEN R S, PEIMANKAR A, and PUTHUSSERYPADY S. A deep learning approach for real-time detection of atrial fibrillation[J]. Expert Systems with Applications, 2019, 115: 465–473. doi: 10.1016/j.eswa.2018.08.011

季虎, 孫即祥, 王春光. 基于小波變換的自適應(yīng)QRS-T對消P波檢測算法[J]. 電子與信息學(xué)報(bào), 2007, 29(8): 1868–1871. doi: 10.3724/SP.J.1146.2006.00117

JI Hu, SUN Jixiang, and WANG Chunguang. An adaptive QRS-T cancellation based on wavelet transform for P-wave detection[J]. Journal of Electronics &Information Technology, 2007, 29(8): 1868–1871. doi: 10.3724/SP.J.1146.2006.00117

PETR?NAS A, S?RNMO L, LUKO?EVI?IUS, et al. Detection of occult paroxysmal atrial fibrillation[J]. Medical & Biological Engineering & Computing, 2015, 53(4): 287–297. doi: 10.1007/s11517-014-1234-y

CUI Xingran, CHANG E, YANG Wenhuang, et al. Automated detection of paroxysmal atrial fibrillation using an information-based similarity approach[J]. Entropy, 2017, 19(12): 677. doi: 10.3390/e19120677

XIN Yi and ZHAO Yizhang. Paroxysmal atrial fibrillation recognition based on multi-scale wavelet α-entropy[J]. BioMedical Engineering OnLine, 2017, 16(1): 121. doi: 10.1186/s12938-017-0406-z

TUZEL O, PORIKLI F, and MEER P. Region covariance: A fast descriptor for detection and classification[C]. Proceedings of the 9th European Conference on Computer Vision, Graz, Austria, 2006: 589–600.

ZHANG Yingying, YANG Cai, and ZHANG Ping. Two-stage sparse coding of region covariance via Log-Euclidean kernels to detect saliency[J]. Neural Networks, 2017, 89: 84–96. doi: 10.1016/j.neunet.2017.02.012

ARSIGNY V, FILLARD P, PENNEC X, et al. Geometric means in a novel vector space structure on symmetric positive-definite matrices[J]. SIAM Journal on Matrix Analysis and Applications, 2007, 29(1): 328–347. doi: 10.1137/050637996

SCH?LKOPF B and SMOLA A J. Learning with Kernels: Support Vector Machines, Regularization, Optimization, and Beyond[M]. Cambridge, USA: MIT Press, 2018: 13. doi: 10.7551/mitpress/4175.003.0018.

LI Peihua, WANG Qilong, ZUO Wangmeng, et al. Log-Euclidean kernels for sparse representation and dictionary learning[C]. 2013 IEEE International Conference on Computer Vision. New York, USA, 2013: 1601–1608. doi: 10.1109/ICCV.2013.202.

GOLDBERGER A L, AMARAl L A N, GLASS L, et al. PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic signals[J]. Circulation, 2000, 101(23): e215–e220. doi: 10.1161/01.CIR.101.23.e215

HARANDI M T, SANDERSON C, HARTLEY R, et al. Sparse coding and dictionary learning for symmetric positive definite matrices: A kernel approach[C]. The 12th European Conference on Computer Vision, Florence, Italy, 2012: 216-229. doi: 10.1007/978-3-642-33709-3_16.

HUANG Chao, YE Shuming, CHEN Hang, et al. A novel method for detection of the transition between atrial fibrillation and sinus rhythm[J]. IEEE Transactions on Biomedical Engineering, 2011, 58(4): 1113–1119. doi: 10.1109/TBME.2010.2096506

ZHOU Xiaolin, DING Hongxia, WU Wanqing, et al. A real-time atrial fibrillation detection algorithm based on the instantaneous state of heart rate[J]. PLoS One, 2015, 10(9): e0136544. doi: 10.1371/journal.pone.0136544

LIAN Jie, WANG Lian, and MUESSIG D. A simple method to detect atrial fibrillation using RR intervals[J]. The American Journal of Cardiology, 2011, 107(10): 1494–1497. doi: 10.1016/j.amjcard.2011.01.028

LEE J, NAM Y, MCMANUS D D, et al. Time-varying coherence function for atrial fibrillation detection[J]. IEEE Transactions on Biomedical Engineering, 2013, 60(10): 2783–2793. doi: 10.1109/TBME.2013.2264721

相關(guān)文章

施引文獻(xiàn)

資源附件(0)

訪問統(tǒng)計(jì)