使用信噪比正则化LightGBM的腹部源胎儿心电信号提取方法

doi:10.12263/DZXB.20211472

电子学报 ›› 2022, Vol. 50 ›› Issue (8): 1793-1800.DOI: 10.12263/DZXB.20211472

• 学术论文 • 下一篇

使用信噪比正则化LightGBM的腹部源胎儿心电信号提取方法

韩亮¹^,², 蔡文涛¹, 蒲秀娟¹^,², 罗统军¹, 黄谦¹

^1.重庆大学微电子与通信工程学院，重庆 400044
^2.生物感知与智能信息处理重庆市重点实验室，重庆 400044

收稿日期:2021-11-01 修回日期:2022-01-25 出版日期:2022-08-25
- 作者简介:
- 韩亮男，1975年生于陕西西乡.生物感知与智能信息处理重庆市重点实验室副主任，重庆大学微电子与通信工程学院副教授.主要研究方向为信号处理和图像处理.E-mail: hanliangaa@cqu.edu.cn
  蔡文涛男，1994年生于四川都江堰.重庆大学微电子与通信工程学院硕士研究生.主要研究方向为生物医学信号处理.E-mail: caiwentao5@cqu.edu.cn
  蒲秀娟女，1979年生于四川隆昌.重庆大学微电子与通信工程学院讲师.主要研究方向为生物医学信号处理.E-mail: puxj@cqu.edu.cn
  罗统军男，1998年生于广西北流.重庆大学微电子与通信工程学院硕士研究生.主要研究方向为生物医学信号处理.E-mail: luotongjun@cqu.edu.cn
  黄谦男，1998年生于四川成都.重庆大学微电子与通信工程学院硕士研究生.主要研究方向为生物医学信号处理.E-mail: Hqian@cqu.edu.cn
- 基金资助:
- 国家自然科学基金 (62171066); 重庆市基础与前沿研究计划项目 (cstc2016jcyjA0376)

Method on Abdominal Electrode-Sourced FECG Extraction Utilizing the Improved LightGBM with Signal-to-Noise Ratio Regularization

HAN Liang¹^,², CAI Wen-tao¹, PU Xiu-juan¹^,², LUO Tong-jun¹, HUANG Qian¹

^1.School of Microelectronics and Communication Engineering，Chongqing University，Chongqing 400044，China
^2.Chongqing Key Laboratory of Bio-perception & Intelligent Information Processing，Chongqing 400044，China

Received:2021-11-01 Revised:2022-01-25 Online:2022-08-25 Published:2022-09-08

摘要/Abstract

摘要：

胎儿心电信号（Fetal ElectroCardioGram，FECG）能反映胎儿健康状况.但是，由于其信噪比相对较低，FECG仍未能在临床上得到广泛应用.如何有效提取高质量的FECG仍是一个巨大挑战.为此，本文提出一种使用信噪比正则化LightGBM（Light Gradient Boosting Machine）模型的FECG提取方法.针对原始母体腹壁混合信号，首先使用传统滤波方法进行噪声抑制，然后再使用快速独立成分分析（Fast Independent Component Analysis，FastICA）从中分离得到母体心电信号（Maternal ElectroCardioGram，MECG）估计和FECG估计，FECG估计中残留的MECG成分是MECG的一种非线性变换.改进传统LightGBM模型，在目标函数中增加FECG的基于互相关系数的信噪比作为正则项，构建信噪比正则化LightGBM模型，并使用该模型拟合这一非线性变换.将MECG估计经由所拟合的非线性变换得到MECG成分的最优估计，并将其抑制，提取得到高质量的FECG.采用真实腹部源心电信号数据集进行实验，结果显示本文提出的方法的灵敏度为99.9%，阳性预测值为99.1%， $F 1$ 分数为99.5%，基于互相关系数与基于特征值分析的信噪比分别为6.0 dB和6.1 dB.实验结果表明，本文提出的方法是有效的且具有更好的性能.

关键词: 胎儿心电信号, FastICA, 基于互相关系数的信噪比, 正则化, LightGBM

Abstract:

The fetal electrocardiogram(FECG) reflects health status of fetus. However, the FECG has not been widely used in clinical practice due to its relatively low signal-to-noise ratio(SNR). How to effectively extract high-quality fetal ECG signal remains a great challenge. In this paper, FECG extraction method utilizing improved LightGBM(light gradient boosting machine) with SNR regularization is proposed. Firstly, the raw maternal abdominal mixed signals are denoised by conventional filtering method. Then, FastICA(fast independent component analysis) is used to separate the maternal electrocardiogram(MECG) estimation and FECG estimation. The residual MECG component in FECG estimation is non-linear transform of MECG and the non-linear transform is fitted by the improved LightGBM with SNR regularization, which is constructed by adding a regularization term, the SNR of extracted FECG, to the objective function of conventional LightGBM. The SNR is estimated using cross correlation. By MECG estimation undergoing the fitted non-linear transform, the residual MECG component in FECG estimation is obtained. At last, the high-quality FECG is extracted by suppressing the estimated MECG component. The real data are adopted to verify the proposed FECG extraction method. The sensitivity, positive predictive value and $F 1$ score of the proposed FECG extraction method are 99.9%, 99.1%, and 99.5%, respectively, and the SNR based on cross correlation and singular value decomposition are 6.0 dB and 6.1 dB, respectively. The experiment results indicate that the proposed method is effective and has better performance.

Key words: fetal electrocardiogram(FECG), FastICA, the SNR based on cross correlation, regularization, LightGBM

中图分类号:

TN911.72

韩亮, 蔡文涛, 蒲秀娟, 罗统军, 黄谦. 使用信噪比正则化LightGBM的腹部源胎儿心电信号提取方法[J]. 电子学报, 2022, 50(8): 1793-1800.

HAN Liang, CAI Wen-tao, PU Xiu-juan, LUO Tong-jun, HUANG Qian. Method on Abdominal Electrode-Sourced FECG Extraction Utilizing the Improved LightGBM with Signal-to-Noise Ratio Regularization[J]. Acta Electronica Sinica, 2022, 50(8): 1793-1800.

图/表 12

图1 FECG采集示意图

图2 使用信噪比正则化LightGBM的FECG提取方法

图3 A15(40~42 s)预处理后数据

图4 A15(40~42 s)经FastICA分解得到的独立成分

表1 信噪比正则化LightGBM模型参数

参数	参数描述	参数值
$J$	时间导数	20
$λ$	目标函数加权系数	-1
max_depth	树的最大深度	7
num_leaves	树的最大叶子数量	31
learning_rate	学习率	0.1
feature_fraction	建树的特征选择比例	0.9

表1 信噪比正则化LightGBM模型参数

参数	参数描述	参数值
$J$	时间导数	20
$λ$	目标函数加权系数	-1
max_depth	树的最大深度	7
num_leaves	树的最大叶子数量	31
learning_rate	学习率	0.1
feature_fraction	建树的特征选择比例	0.9

图5 A15 (40~42 s)经FastICA-信噪比正则化LightGBM提取的FECG

表2 使用FastICA的FECG提取方法的性能评价

数据集	$S N R c o r$ /dB	$S N R s v d$ /dB	Se/%	PPV/%	$F 1$ /%
A04	1.8	2.1	100	78.1	87.7
A08	3.2	3.4	99.2	69.2	81.6
A15	2.0	2.4	97.0	81.1	88.4
A22	0.9	1.2	96.8	96.0	96.4
A32	4.3	5.0	98.7	96.8	97.7
A43	5.4	5.7	100	98.1	99.1
A59	4.0	4.2	98.7	98.7	98.7
平均值	3.1	3.4	98.6	88.3	92.8

表2 使用FastICA的FECG提取方法的性能评价

数据集	$S N R c o r$ /dB	$S N R s v d$ /dB	Se/%	PPV/%	$F 1$ /%
A04	1.8	2.1	100	78.1	87.7
A08	3.2	3.4	99.2	69.2	81.6
A15	2.0	2.4	97.0	81.1	88.4
A22	0.9	1.2	96.8	96.0	96.4
A32	4.3	5.0	98.7	96.8	97.7
A43	5.4	5.7	100	98.1	99.1
A59	4.0	4.2	98.7	98.7	98.7
平均值	3.1	3.4	98.6	88.3	92.8

表3 使用FastICA-信噪比正则化LightGBM的FECG提取方法的性能评价

数据集	$S N R c o r$ /dB	$S N R s v d$ /dB	Se/%	PPV/%	$F 1$ /%
A04	4.4	4.5	100	99.2	99.6
A08	5.8	6.0	100	98.5	99.2
A15	7.2	7.3	100	99.3	99.6
A22	5.3	5.4	100	99.2	99.6
A32	6.7	6.8	100	99.3	99.7
A43	7.1	7.3	100	99.4	99.7
A59	5.5	5.6	99.3	98.7	99.0
平均值	6.0	6.1	99.9	99.1	99.5

表3 使用FastICA-信噪比正则化LightGBM的FECG提取方法的性能评价

数据集	$S N R c o r$ /dB	$S N R s v d$ /dB	Se/%	PPV/%	$F 1$ /%
A04	4.4	4.5	100	99.2	99.6
A08	5.8	6.0	100	98.5	99.2
A15	7.2	7.3	100	99.3	99.6
A22	5.3	5.4	100	99.2	99.6
A32	6.7	6.8	100	99.3	99.7
A43	7.1	7.3	100	99.4	99.7
A59	5.5	5.6	99.3	98.7	99.0
平均值	6.0	6.1	99.9	99.1	99.5

图6 A15(40~42 s)对比实验1

图7 A15(40~42 s)对比实验2

图8 A15 (40~42 s)对比实验3

表4 FECG提取方法性能对比(平均值)

方法	$S N R c o r$ /dB	$S N R s v d$ /dB	Se/%	PPV/%	$F 1$ /%
TFBSS	2.0	2.3	94.5	77.8	84.2
RobustICA	2.1	2.8	93.1	81.0	86.3
FastICA	3.1	3.4	98.6	88.3	92.8
TFBSS-LightGBM	2.8	3.2	99.5	98.6	99.0
RobustICA-LightGBM	4.9	5.0	99.7	98.9	99.3
FastICA-LightGBM	5.1	5.3	99.7	98.9	99.3
TFBSS-SNR regularization LightGBM	4.5	4.7	99.5	97.1	98.2
RobustICA-SNR regularization LightGBM	5.9	6.0	99.9	99.0	99.5
FastICA-SNR regularization LightGBM	6.0	6.1	99.9	99.1	99.5

表4 FECG提取方法性能对比(平均值)

方法	$S N R c o r$ /dB	$S N R s v d$ /dB	Se/%	PPV/%	$F 1$ /%
TFBSS	2.0	2.3	94.5	77.8	84.2
RobustICA	2.1	2.8	93.1	81.0	86.3
FastICA	3.1	3.4	98.6	88.3	92.8
TFBSS-LightGBM	2.8	3.2	99.5	98.6	99.0
RobustICA-LightGBM	4.9	5.0	99.7	98.9	99.3
FastICA-LightGBM	5.1	5.3	99.7	98.9	99.3
TFBSS-SNR regularization LightGBM	4.5	4.7	99.5	97.1	98.2
RobustICA-SNR regularization LightGBM	5.9	6.0	99.9	99.0	99.5
FastICA-SNR regularization LightGBM	6.0	6.1	99.9	99.1	99.5

参考文献 22

1	MARTINEK R, KAHANKOVA R, JEZEWSKI J, et al. Comparative effectiveness of ICA and PCA in extraction of fetal ECG from abdominal signals: Toward non-invasive fetal monitoring[J]. Frontiers in Physiology, 2018, 9: 648.
2	BEHAR J, ANDREOTTI F, ZAUNSEDER S, et al. A practical Guide to non-invasive foetal electrocardiogram extraction and analysis[J]. Physiological Measurement, 2016, 37(5): R1-R35.
3	JEZEWSKI J, WROBEL J, MATONIA A, et al. Is abdominal fetal electrocardiography an alternative to Doppler ultrasound for FHR variability evaluation?[J]. Frontiers in Physiology, 2017, 8: 305.
4	TAHA L Y, ABDEL-RAHEEM E. Fetal ECG extraction using input-mode and output-mode adaptive filters with blind source separation[J]. Canadian Journal of Electrical and Computer Engineering, 2020, 43(4): 295-304.
5	WARMERDAM G J J, VULLINGS R, SCHMITT L, et al. Hierarchical probabilistic framework for fetal R-peak detection, using ECG waveform and heart rate information[J]. IEEE Transactions on Signal Processing, 2018, 66(16): 4388-4397.
6	KAHANKOVA R, MARTINEK R, JAROS R, et al. A review of signal processing techniques for non-invasive fetal electrocardiography[J]. IEEE Reviews in Biomedical Engineering, 2020, 13:51-73.
7	KAHANKOVA R, MARTINEK R, BILIK P. Non-invasive fetal ECG extraction from maternal abdominal ECG using LMS and RLS adaptive algorithms[C]//Proceedings of the Third International Afro-European Conference for Industrial Advancement(AECIA 2016). Marrakesh: Springer, 2016. 258-271.
8	ASSALEH K. Extraction of fetal electrocardiogram using adaptive neuro-fuzzy inference systems[J]. IEEE Transactions on Biomedical Engineering, 2007, 54(1): 59-68.
9	HAN L, PU XIU-JUAN, CHEN XIAO-JUN. Method of fetal electrocardiogram extraction based on ν-support vector regression[J]. IET Signal Processing, 2015, 9(5): 430-439.
10	TSUI S Y, LIU C S, LIN C W. Modified maternal ECG cancellation for portable fetal heart rate monitor[J]. Biomedical Signal Processing and Control, 2017, 32: 76-81.
11	韩亮, 蔡文涛, 蒲秀娟, 龙灵. 结合FastICA与EKF的腹部源胎儿心电信号提取[J]. 仪器仪表学报, 2021, 42(7):116-125.
	HAN L, CAI W T, PU X J, LONG L. Abdominal electrode-sourced FECG extraction utilizing EKF combined with FastICA[J]. Chinese Journal of Scientific Instrument, 2021, 42(7):116-125. (in Chinese)
12	LIMA-HERRERA S L, ALVARADO-SERRANO C, HERNÁNDEZ-RODRÍGUEZ P R. Fetal ECG extraction based on adaptive filters and Wavelet Transform: Validation and application in fetal heart rate variability analysis[C]//2016 13th International Conference on Electrical Engineering, Computing Science and Automatic Control(CCE). Mexico City: IEEE, 2016: 1-6.
13	ANDREOTTI F, RIEDL M, HIMMELSBACH T, et al. Robust fetal ECG extraction and detection from abdominal leads[J]. Physiological Measurement, 2014, 35(8): 1551-1567.
14	KE G, MENG Q, FINLEY T, et al. Lightgbm: A highly efficient gradient boosting decision tree[C]//Advances in Neural Information Processing Systems 30(NIPS 2017). Long Beach: Curran Associates, Inc., 2017: 1-9.
15	孙全明, 曲志坚, 任崇广. 基于粒子群优化和LightGBM的情景感知多式联运推荐[J]. 电子学报, 2021, 49(5): 894-903.
	SUN Q M, QU Z J, REN C G. Context-aware multi-modal transportation recommendation based on particle swarm optimization and LightGBM[J]. Acta Electronica Sinica, 2021, 49(5): 894-903. (in Chinese)
16	Outram N J. Intelligent Pattern Analysis of the Foetal Electrocardiogram[D]. Plymouth: University of Plymouth, 1997.
17	VARANINI M, TARTARISCO G, BILLECI L, et al. A multi-step approach for non-invasive fetal ECG analysis[C]//Computing in Cardiology Conference. Zaragoza: IEEE, 2013: 281-284.
18	MARTINEZ M, CALPE J, SORIA E, et al. Methods to evaluate the performance of fetal electrocardiogram extraction algorithms[C]//Computers in Cardiology. Rotterdam: IEEE, 2001: 253-256.
19	BEHAR J, OSTER J, CLIFFORD G D. Combining and benchmarking methods of foetal ECG extraction without maternal or scalp electrode data[J]. Physiological Measurement, 2014, 35(8): 1569-1589.
20	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): E215-E220.
21	JOHNSON A E W, BEHAR J, ANDREOTTI F, et al. Multimodal heart beat detection using signal quality indices[J]. Physiological Measurement, 2015, 36(8): 1665-1677.
22	张昱, 刘开峰, 张全新, 王艳歌, 高凯龙. 基于组合-卷积神经网络的中文新闻文本分类[J]. 电子学报, 2021, 49(6): 1059-1067.
	ZHANG Y, LIU K F, ZHANG Q X, WANG K G, GAO K L. A combined-convolutional neural network for Chinese news text classification[J]. Acta Electronica Sinica, 2021, 49(6): 1059-1067. (in Chinese)

[1]	江兵兵, 何文达, 吴兴宇, 项俊浩, 洪立斌, 盛伟国. 基于自适应图学习的半监督特征选择[J]. 电子学报, 2022, 50(7): 1643-1652.
[2]	杨晓城, 杨真乙, 阎敬业, 武林, 蒋明峰. 基于拉普拉斯算子的综合孔径辐射计成像算法[J]. 电子学报, 2022, 50(2): 339-345.
[3]	刘浩阳, 林耀进, 刘景华, 吴镒潾, 毛煜, 李绍滋. 由粗到细的分层特征选择[J]. 电子学报, 2022, 50(11): 2778-2789.
[4]	仇祝令, 查宇飞, 吴敏, 王青. 基于注意力学习的正则化相关滤波跟踪算法[J]. 电子学报, 2020, 48(9): 1762-1768.
[5]	李东瑾, 杨瑞娟, 李晓柏, 董睿杰. 基于栈式稀疏降噪自编码网络的辐射源调制识别[J]. 电子学报, 2020, 48(6): 1198-1204.
[6]	吉利鹏, 倪锦根. 偏差补偿符号子带自适应滤波器[J]. 电子学报, 2020, 48(11): 2220-2225.
[7]	蒲磊, 冯新喜, 侯志强, 查宇飞, 余旺盛. 基于深度空间正则化的相关滤波跟踪算法[J]. 电子学报, 2020, 48(10): 2025-2032.
[8]	任开旭, 王玉龙, 刘同存, 李炜. 融合多维语义表示的概率矩阵分解模型[J]. 电子学报, 2019, 47(9): 1848-1854.
[9]	骆睿, 刘莉, 佟瑞, 李凌云. 基于Tikhonov正则化的高分辨率群时延测量与计算方法[J]. 电子学报, 2019, 47(5): 1044-1048.
[10]	李亚峰. 一种基于多字典学习的图像分割模糊方法[J]. 电子学报, 2018, 46(7): 1700-1709.
[11]	陈兵飞, 江兵兵, 周熙人, 陈欢欢. 基于稀疏贝叶斯的流形学习[J]. 电子学报, 2018, 46(1): 98-103.
[12]	闫震海, 杨飞然, 杨军. 一种变正则化矩阵的改进多带结构子带自适应滤波算法[J]. 电子学报, 2017, 45(8): 2019-2025.
[13]	卜红霞, 白霞, 赵娟, 齐耀辉, 闫若颖. 基于压缩感知的矩阵型联合SAR成像与自聚焦算法[J]. 电子学报, 2017, 45(4): 874-881.
[14]	于海平, 何发智, 潘一腾, 陈晓. 一种基于多特征的距离正则化水平集快速分割方法[J]. 电子学报, 2017, 45(3): 534-539.
[15]	张立峰, 刘昭麟, 田沛. 基于压缩感知的电容层析成像图像重建算法[J]. 电子学报, 2017, 45(2): 353-358.

使用信噪比正则化LightGBM的腹部源胎儿心电信号提取方法

Method on Abdominal Electrode-Sourced FECG Extraction Utilizing the Improved LightGBM with Signal-to-Noise Ratio Regularization

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 12

参考文献 22

相关文章 15

编辑推荐

Metrics