Limits...
Powerline noise elimination in biomedical signals via blind source separation and wavelet analysis.

Akwei-Sekyere S - PeerJ (2015)

Bottom Line: However, due to the instability of biomedical signals, the distribution of signals filtered out may not be centered at 50/60 Hz.Since powerline noise is additive in nature, it is intuitive to model powerline noise in a raw recording and subtract it from the raw data in order to obtain a relatively clean signal.The performance of this algorithm was compared with that of a 4th order band-stop Butterworth filter, empirical mode decomposition, independent component analysis and, a combination of empirical mode decomposition with independent component analysis.

View Article: PubMed Central - HTML - PubMed

Affiliation: Neuroscience Program, Michigan State University , East Lansing, MI , USA.

ABSTRACT
The distortion of biomedical signals by powerline noise from recording biomedical devices has the potential to reduce the quality and convolute the interpretations of the data. Usually, powerline noise in biomedical recordings are extinguished via band-stop filters. However, due to the instability of biomedical signals, the distribution of signals filtered out may not be centered at 50/60 Hz. As a result, self-correction methods are needed to optimize the performance of these filters. Since powerline noise is additive in nature, it is intuitive to model powerline noise in a raw recording and subtract it from the raw data in order to obtain a relatively clean signal. This paper proposes a method that utilizes this approach by decomposing the recorded signal and extracting powerline noise via blind source separation and wavelet analysis. The performance of this algorithm was compared with that of a 4th order band-stop Butterworth filter, empirical mode decomposition, independent component analysis and, a combination of empirical mode decomposition with independent component analysis. The proposed method was able to expel sinusoidal signals within powerline noise frequency range with higher fidelity in comparison with the mentioned techniques, especially at low signal-to-noise ratio.

No MeSH data available.


Related in: MedlinePlus

Independent component selection, signal reconstruction and the effect of time window.(A) The blue, green and red signals are the recovered, original and noised signals respectively. (B) The blue, green and red signals are the power spectrum for the recovered, original and noised signals respectively. (C) The grey trace is the AC noise extracted and the black trace is the Manhattan distance between the original signal and the recovered signal. (D) The upper image is the pseudo-convolution of the selected independent component for the first 200 ms of the data. The lower represents that of the final 50 ms. Note that they both peak at circa 60 Hz.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4493666&req=5

fig-2: Independent component selection, signal reconstruction and the effect of time window.(A) The blue, green and red signals are the recovered, original and noised signals respectively. (B) The blue, green and red signals are the power spectrum for the recovered, original and noised signals respectively. (C) The grey trace is the AC noise extracted and the black trace is the Manhattan distance between the original signal and the recovered signal. (D) The upper image is the pseudo-convolution of the selected independent component for the first 200 ms of the data. The lower represents that of the final 50 ms. Note that they both peak at circa 60 Hz.

Mentions: To demonstrate the viability of the process of powerline noise removal, artificial 60 Hz noise was added to local field potentials (SNR = 4.2816dB; amplitude = 700) and was subsequently reconstructed. A 200 ms time window (without overlap) was used to denoise 250 ms of the mentioned neural data. As shown in Figs. 2A and 2C, the procedure was able to recover a sufficient amount of the original signal with little Manhattan distances. The morphology of the Manhattan distance suggests that the procedure extracted the appropriate frequency, but obtained relatively erroneous amplitude information. As shown in Figs. 2B and 2D, the pseudo-convolution procedure for extraction of powerline noise chose the 60 Hz noise—which is the right frequency for the noise model in this context. Since the pseudo-convolution between the extracted alternating current noise peaks at about 60 Hz on all translations, it implies that the summation of the pseudo-convolution values on every translation should result in a function that also peaks at about 60 Hz. Thus, whichever independent component that has its summed pseudo-convolution peak closest to 60 Hz is indeed the desired powerline noise. Although the final round of blind-source separation in this algorithm requires prior knowledge about the statistical properties of the source data being approximated, it is plausible that this constraint is not a hindrance but, instead, facilitates the process. With the assumption of statistical independence, at least one of the resulting approximations of the source data is forced to look undeniably unique. Signals of this form are usually some fluctuations that are chronically present in the mixed data. For this reason, ICA effectively serves as a helper for the extraction and identification of the desired powerline noise to be removed. The Manhattan distance between the original signal and the denoised signal increased during the final 50 ms of the neural data. This suggests the approach does not provide good results with very small time windows. Essentially, the best approach will be to run the algorithm without a time window. In conclusion, it is most appropriate for offline signal analysis.


Powerline noise elimination in biomedical signals via blind source separation and wavelet analysis.

Akwei-Sekyere S - PeerJ (2015)

Independent component selection, signal reconstruction and the effect of time window.(A) The blue, green and red signals are the recovered, original and noised signals respectively. (B) The blue, green and red signals are the power spectrum for the recovered, original and noised signals respectively. (C) The grey trace is the AC noise extracted and the black trace is the Manhattan distance between the original signal and the recovered signal. (D) The upper image is the pseudo-convolution of the selected independent component for the first 200 ms of the data. The lower represents that of the final 50 ms. Note that they both peak at circa 60 Hz.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4493666&req=5

fig-2: Independent component selection, signal reconstruction and the effect of time window.(A) The blue, green and red signals are the recovered, original and noised signals respectively. (B) The blue, green and red signals are the power spectrum for the recovered, original and noised signals respectively. (C) The grey trace is the AC noise extracted and the black trace is the Manhattan distance between the original signal and the recovered signal. (D) The upper image is the pseudo-convolution of the selected independent component for the first 200 ms of the data. The lower represents that of the final 50 ms. Note that they both peak at circa 60 Hz.
Mentions: To demonstrate the viability of the process of powerline noise removal, artificial 60 Hz noise was added to local field potentials (SNR = 4.2816dB; amplitude = 700) and was subsequently reconstructed. A 200 ms time window (without overlap) was used to denoise 250 ms of the mentioned neural data. As shown in Figs. 2A and 2C, the procedure was able to recover a sufficient amount of the original signal with little Manhattan distances. The morphology of the Manhattan distance suggests that the procedure extracted the appropriate frequency, but obtained relatively erroneous amplitude information. As shown in Figs. 2B and 2D, the pseudo-convolution procedure for extraction of powerline noise chose the 60 Hz noise—which is the right frequency for the noise model in this context. Since the pseudo-convolution between the extracted alternating current noise peaks at about 60 Hz on all translations, it implies that the summation of the pseudo-convolution values on every translation should result in a function that also peaks at about 60 Hz. Thus, whichever independent component that has its summed pseudo-convolution peak closest to 60 Hz is indeed the desired powerline noise. Although the final round of blind-source separation in this algorithm requires prior knowledge about the statistical properties of the source data being approximated, it is plausible that this constraint is not a hindrance but, instead, facilitates the process. With the assumption of statistical independence, at least one of the resulting approximations of the source data is forced to look undeniably unique. Signals of this form are usually some fluctuations that are chronically present in the mixed data. For this reason, ICA effectively serves as a helper for the extraction and identification of the desired powerline noise to be removed. The Manhattan distance between the original signal and the denoised signal increased during the final 50 ms of the neural data. This suggests the approach does not provide good results with very small time windows. Essentially, the best approach will be to run the algorithm without a time window. In conclusion, it is most appropriate for offline signal analysis.

Bottom Line: However, due to the instability of biomedical signals, the distribution of signals filtered out may not be centered at 50/60 Hz.Since powerline noise is additive in nature, it is intuitive to model powerline noise in a raw recording and subtract it from the raw data in order to obtain a relatively clean signal.The performance of this algorithm was compared with that of a 4th order band-stop Butterworth filter, empirical mode decomposition, independent component analysis and, a combination of empirical mode decomposition with independent component analysis.

View Article: PubMed Central - HTML - PubMed

Affiliation: Neuroscience Program, Michigan State University , East Lansing, MI , USA.

ABSTRACT
The distortion of biomedical signals by powerline noise from recording biomedical devices has the potential to reduce the quality and convolute the interpretations of the data. Usually, powerline noise in biomedical recordings are extinguished via band-stop filters. However, due to the instability of biomedical signals, the distribution of signals filtered out may not be centered at 50/60 Hz. As a result, self-correction methods are needed to optimize the performance of these filters. Since powerline noise is additive in nature, it is intuitive to model powerline noise in a raw recording and subtract it from the raw data in order to obtain a relatively clean signal. This paper proposes a method that utilizes this approach by decomposing the recorded signal and extracting powerline noise via blind source separation and wavelet analysis. The performance of this algorithm was compared with that of a 4th order band-stop Butterworth filter, empirical mode decomposition, independent component analysis and, a combination of empirical mode decomposition with independent component analysis. The proposed method was able to expel sinusoidal signals within powerline noise frequency range with higher fidelity in comparison with the mentioned techniques, especially at low signal-to-noise ratio.

No MeSH data available.


Related in: MedlinePlus