An overview of methods for eliminating noise from

ECG signals

Abstract—This study investigates the denoising of electrocar- and interval properties (PR-segment, ST-segment, PR interval,
diogram (ECG) signals to increase the accuracy of detecting ST interval, RR interval, etc.). These features must be correctly
heart conditions. The study categorizes and assesses the various determined for ECG signal-denoising applications. Due to
denoising methods researchers propose using different ECG
databases and benchmark metrics. Wavelet-VBE, EMD-MAF, multiple prevalent disturbances, denoising ECG signals is
GAN2, GSSSA, new MP-EKF, DLSR, and AKF perform better difficult but essential for appropriate cardiac disease diagnosis
for removing additive white Gaussian noise than GAN1 for and therapy, according to the study. [8], [9]
removing baseline wander and electrode motion artifacts. DLSR
and EWT perform well for power-line interference elimination. II. L ITERATURE R EVIEW
FCN-based DAE, DWT (Sym6) soft, MABWT (soft), CPSD
sparsity, and UWT are promising for composite noise removal. Due to maternal and fetal ECG’s comparable spatial and
The study concludes that these denoising techniques can improve temporal content, (John et al., 2019) propose a simple yet ef-
ECG signal analysis and aid in detecting cardiac conditions. fective method for extracting fetal ECG from abdominal ECG.
Additionally, a 1D digital ECG signal is extracted to access the
I. INTRODUCTION
exact values of the extracted points. Automatic digital ECG
This study examines the denoising of electrocardiogram signal extraction and standard QRS recognition from real-
(ECG) signals to improve the accuracy of detecting heart con- scene ECG images are investigated (Wang et al., 2019). In light
ditions. The study categorizes and evaluates various denois- of these responsibilities, we propose an automatic digital ECG
ing methods proposed by researchers on benchmark metrics signal extraction and standard QRS recognition from ECG
using different ECG databases. [1] This study investigates images of real-world scenes. The input signal will be filtered
the denoising of electrocardiogram (ECG) signals to increase with two filters (FIR filter with window method and IIR filter
the accuracy of detecting heart conditions. The study cate- with pole-zero placement method or bilinear transformation
gorizes and evaluates various denoising methods researchers method). Their ability to eliminate or reduce the noise signal
propose using different ECG databases as benchmark metrics. will be compared (Leonard et al., 2020). (Qiu et al., 2020)
[2] DLSR and EWT remove power-line interference. Com- investigate a deep convolutional network-based two-stage ECG
posite noise removal is promising using FCN-based DAE, signal denoising method. A noise reduction technique based
DWT (Sym6) soft, MABWT, CPSD sparsity, and UWT. The on deep learning is proposed. Modak et al. (2021) present a
study found that denoising approaches improve ECG signal new algorithm for peak detection that can suppress noise and
processing and cardiac disease detection. Electrocardiograms adapt to changes in the morphology of the ECG signal for
(ECGs) show ionic current flow and heart electrical and muscle improved detection performance. Pashko et al. (2002) used
activities. Heart disease diagnosis relies on physiological and various filters to eliminate ECG baseline drift: low-pass filters
pathological data. [3] [4] Cardio-respiratory monitoring and based on the forward and inverse discrete Fourier transform,
seizure detection use ECG monitoring. Base-line drift, mus- Butterworth filter, median filter, and Savitsky-Golay filter.
cular artifacts, and power-line interference make it difficult to Modak et al. (2021) present a new method of QRS detection
detect morphological defects using ECG. ECG signals have using advanced adaptive multilevel thresholding (AAMT) with
morphological (P-wave, QRS-complex, T-wave, and U-wave) selective statistical false peak elimination (SSFPE). Deep
and interval properties (PR-segment, ST-segment, PR interval, convolutional network-based two-stage ECG signal denoising
ST interval, RR interval, etc.). [5] [6] These features must be is studied (Qiu et al., 2021). U-net + DR-net can obtain the best
correctly determined for ECG signal-denoising applications. SNR, RMSE, and indicator performance compared to the other
Due to multiple prevalent disturbances, denoising ECG signals five schemes (FCN, U-net, etc.). (Xu et al., 2021) investigate
is difficult but essential for appropriate cardiac disease diag- accurate heart rate and respiration rate detection using non-
nosis and therapy, according to the study. Electrocardiograms contact radar sensors and a peak selection method based
(ECGs) show ionic current flow and heart electrical and muscle on higher-order harmonics. A signal-processing technique for
activities. Heart disease diagnosis relies on physiological and UWB radar systems was developed to detect the heart rate and
pathological data. Cardiorespiratory monitoring and seizure respiration rate.
detection use ECG monitoring. [7] Base-line drift, muscu-
lar artifacts, and power-line interference make it difficult to R EFERENCES
detect morphological defects using ECG. ECG signals have [1] Li, W, (2019) “Wavelets for electrocardiogram:overview and taxonomy’,
morphological (P-wave, QRS-complex, T-wave, and U-wave) 7”, pp. 25627–25649, IEEE Access 2019
[2] He, R., Wang, K., Li, Q., et al, (2017) “A novel method for the detection ”Automatic Digital ECG Signal Extraction and Normal QRS Recogni-
of R-peaks in ECG based on K-nearest neighbors and particle swarm tion from Real Scene ECG Images”, COMPUTER METHODS AND
optimization” , EURASIP J. Adv. Signal Process., 2017, pp. 1–14. PROGRAMS IN BIOMEDICINE, 2019. (IF: 3)
[3] Lastre-Dominguez, C., Shmaliy, Y.S., Ibarra-Manzano, O., et al, (2018) [12] Christopher Leonard; Nabila Husna Shabrina; ”Analisis Keefektifan
”Fiducial features extraction for ECG signals using state-space unbi- Penggunaan Filter FIR Dan IIR Pada Sinyal Pernapasan EMGdi Dengan
ased FIR smoothing” ,Autumn Meeting Power, Electronics Computing Simulasi MATLAB”, 2020.
(ROPEC), Ixtapa, Mexico, pp. 1–6, IEEE Int. [13] L. Qiu; W. Cai; J. Yu; J. Zhong; Y. Wang; W. Li; Y. Chen; L. Wang; ”A
[4] Kang, S.J., Lee, S.Y., Cho, H.I., et al, (2016) ”ECG authentication system Two-stage ECG Signal Denoising Method Based On Deep Convolutional
design based on signal analysis in mobile and wearable devices” , IEEE Network”, BIO.BIOENGINEERING, 2020.
Signal Process. Lett., 2016, 23, (6), pp. 805–808 [14] Sudipta Modak; Luay Yassin Taha; Esam Abdel-Raheem; ”A Novel
[5] Lee, W.K., Yoon, H., Park, K.S. (2016) ”Smart ECG monitoring patch Method of QRS Detection Using Time and Amplitude Thresholds With
with built-in R-peak detection for long-term HRV analysis” , Ann. Statistical False Peak Elimination”, IEEE ACCESS, 2021. (IF: 3)
Biomed. Eng., 44, (7), pp. 2292–2301 [15] Anatolii Pashko; Iurii Krak; Oleg Stelia; Waldemar Wojcik; ”Base-
[6] Friesen, G.M., Jannett, T.C., Jadallah, M.A., et al. , (1990) ”A comparison line Wander Correction Of the Electrocardiogram Signals For Effective
of the noise sensitivity of nine QRS detection algorithms”, IEEE Trans. Preprocessing”, LECTURE NOTES IN COMPUTATIONAL INTELLI-
Biomed. Eng., 37, (1), pp. 85–98 GENCE AND DECISION ..., 2021.
[7] Satija, U., Ramkumar, B., Sabarimalai Manikandan, M., (2018) ”A review [16] Sudipta Modak; Esam Abdel-Raheem; Luay Yassin Taha; ”A Novel
of signal processing techniques for electrocardiogram signal quality Adaptive Multilevel Thresholding Based Algorithm for QRS Detection”,
assessment”, IEEE Rev. Biomed. Eng., 11, pp. 36–52 BIOMEDICAL ENGINEERING ADVANCES, 2021.
[8] Van Alsté, J., Schilder, T., (1985) ”Removal of base-line wander and [17] S. T. Sanamdikar; S. T. Hamde; V. G. Asutkar; ”Classification and
power-line interference from the ECG by an efficient FIR filter with Analysis of Cardiac Arrhythmia Based on Incremental Support Vector Re-
a reduced number of taps” , IEEE Trans. Biomed. Eng., 32, (12), pp. gression on IOT Platform”, BIOMED. SIGNAL PROCESS. CONTROL.,
1052–1060 2021.
[9] Levkov, C., Mihov, G., Ivanov, R., et al. , (2005) ”Removal of power- [18] Lishen Qiu; Wenqiang Cai; Miao Zhang; Wenliang Zhu; Lirong Wang;
line interference from the ECG: a review of the subtraction procedure”, ”Two-stage ECG Signal Denoising Based on Deep Convolutional Net-
Biomed. Eng. Online, 50, (4), pp. 1–18 work”, PHYSIOLOGICAL MEASUREMENT, 2021.
[10] Rolant Gini John; K I Ramachandran; ”Extraction Of Foetal ECG [19] Hongqiang Xu; Malikeh P Ebrahim; Kareeb Hasan; Fatemeh Heydari;
From Abdominal ECG By Nonlinear Transformation And Estimations”, Paul Howley; Mehmet Rasit Yuce; ”Accurate Heart Rate and Respira-
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE, 2019. tion Rate Detection Based on A Higher-Order Harmonics Peak Selec-
(IF: 3) tion Method Using Radar Non-Contact Sensors”, SENSORS (BASEL,
[11] Shuang Wang; Shugang Zhang; Zhen Li; Lei Huang; Zhiqiang Wei; SWITZERLAND), 2021. (IF: 3)