Identifying Noisy ECG Signals in Large Datasets Using a Temporal Convolutional Neural Network Trained to Estimate Pseudo-SNR
- Background: Electrocardiogram (ECG) signals are often contaminated by noise. Manual review of large ECG databases to identify noisy signals is time-consuming. Traditional signal quality assessment algorithms often do not generalize well or are computationally expensive. This study developed a Temporal Convolutional Neural Network (TCNN) to estimate the signal-to-noise ratio (SNR) of ECG signals. Method: We trained a TCNN on a proprietary database of 134,019 12-lead ECGs without any machine or human-added noise labels. Assuming that this data had high SNR, we randomly selected a single lead from each ECG and added random Gaussian noise. We then scaled the signals and added noise to give a negatively skewed normal distribution of true SNR values. We trained a TCNN to regress low- and high-frequency pseudo-SNR values from the raw noisy input signals. Results: On the testing dataset, the TCNN achieved a mean error of 0.31±1.80 dB and a Pearson correlation coefficient of 0.96 for low-frequency pseudo-SNR. Similarly, for high-frequencyBackground: Electrocardiogram (ECG) signals are often contaminated by noise. Manual review of large ECG databases to identify noisy signals is time-consuming. Traditional signal quality assessment algorithms often do not generalize well or are computationally expensive. This study developed a Temporal Convolutional Neural Network (TCNN) to estimate the signal-to-noise ratio (SNR) of ECG signals. Method: We trained a TCNN on a proprietary database of 134,019 12-lead ECGs without any machine or human-added noise labels. Assuming that this data had high SNR, we randomly selected a single lead from each ECG and added random Gaussian noise. We then scaled the signals and added noise to give a negatively skewed normal distribution of true SNR values. We trained a TCNN to regress low- and high-frequency pseudo-SNR values from the raw noisy input signals. Results: On the testing dataset, the TCNN achieved a mean error of 0.31±1.80 dB and a Pearson correlation coefficient of 0.96 for low-frequency pseudo-SNR. Similarly, for high-frequency pseudo-SNR, the mean error was 0.29±1.63 dB and the Pearson correlation coefficient was 0.97. Conclusion: A Temporal Convolutional Neural Network can accurately estimate the SNR of unseen ECGs.…
Author: | Daniel GüldenringORCiD, Peter DoggartORCiD, Alan Kennedy, Raymond BondORCiD, Dewar FinlayORCiD |
---|---|
DOI: | https://doi.org/10.22489/CinC.2023.103 |
Identifier: | 979-8-3503-8252-5 OPAC HS OPAC extern |
Identifier: | 2325-887X OPAC HS OPAC extern |
Parent Title (English): | Computing in Cardiology 2023 |
Publisher: | IEEE |
Place of publication: | Piscataway |
Document Type: | conference proceeding (article) |
Conference: | 2023 Computing in Cardiology (CinC) 01-04. Oct. 2023, Atlanta, Georgia, USA |
Language: | English |
Date of Publication (online): | 2024/02/16 |
Year of first Publication: | 2023 |
Tag: | Biomedizinische Signalverarbeitung ECG |
GND Keyword: | Künstliche Intelligenz; Elektrokardiogramm |
Volume: | 2023 |
Issue: | 50 |
Number of pages: | 4 Seiten |
First Page: | 1 |
Last Page: | 4 |
Institutes: | Fakultät Elektrotechnik |
Dewey Decimal Classification: | 0 Informatik, Informationswissenschaft, allgemeine Werke / 00 Informatik, Wissen, Systeme / 000 Informatik, Informationswissenschaft, allgemeine Werke |
Open Access: | open_access |
Research focus: | Eigenständige Forschungsfelder |
Publication Lists: | Güldenring, Daniel |
Publication reviewed: | begutachtet |
Licence (German): | Es gilt das deutsche Urheberrecht |
Release Date: | 2024/02/16 |