Frequency- and bit-rate-adaptive compression of electromyography signals by wavelet packet transform and optimal scalar quantizer bit rate allocation
Artificial Intelligence in Health · 23.06.2026 · 026140011
DOI: 10.36922/nsce026140011 ↗Kurz gefasst
Fachartikel in der Zeitschrift Artificial Intelligence in Health, online erschienen am 23.06.2026. Verfasst von Joseph Mvogo Ngono und 7 weiteren Autor:innen.
Abstract (Originalsprache)
Electromyography (EMG) signals are widely used for neuromuscular assessment, rehabilitation, and prosthetic control, but their high sampling rates and long recording durations generate large data volumes. This study presents a frequency- and bit-rate-adaptive compression method for EMG signals based on wavelet packet transform and optimal scalar quantizer bit allocation. The method first decomposed the signal using a complete wavelet packet tree of depth L = 5. Then, the coefficients of each sub-band were modeled as generalized Gaussian distributions, and analytical rate-distortion estimates were derived for a uniform scalar quantizer. These estimates were used in a Lagrangian optimization framework to jointly select the optimal wavelet packet basis and allocate quantization bit rates to achieve the target compression ratio. Experimentswere conducted on four EMG records, including two surface EMG signals and two needle EMG signals, using Daubechies db2 and biorthogonal bior4.4 wavelets over compression ratios from 5 to 25. The proposed method was compared with a conventional wavelet-based baseline using root-meansquare error (RMSE), signal-to-noise ratio (SNR), percentage root-mean-square difference (PRD), and quality score (QS). Across all tested operating points, the optimal adaptive configuration reduced RMSE and PRD by 37.7% on average, increased SNR by 4.33 dB, and improved QS by 69.5% compared with the optimal baseline configuration. These results highlight the effectiveness of the proposed adaptive framework, making it well-suited for real-time biomedical signal compression and telemedicine applications.
Zitieren
APA
Ngono, J. M., Nanfak, A., Nkapkop, J. D. D., Woguia, A. N., Etoundi, C. L., Etoundi, C. L., Njitacke, Z. T., & Ele, P. (2026). Frequency- and bit-rate-adaptive compression of electromyography signals by wavelet packet transform and optimal scalar quantizer bit rate allocation. Artificial Intelligence in Health, 0(0), 026140011. https://doi.org/10.36922/nsce026140011
Vancouver
Ngono JM, Nanfak A, Nkapkop JDD, Woguia AN, Etoundi CL, Etoundi CL, et al. Frequency- and bit-rate-adaptive compression of electromyography signals by wavelet packet transform and optimal scalar quantizer bit rate allocation. Artificial Intelligence in Health. 2026;0(0):026140011. doi:10.36922/nsce026140011
BibTeX
@article{Ngono2026,
author = {Ngono, Joseph Mvogo and Nanfak, Arnaud and Nkapkop, Jean de Dieu and Woguia, Auguste Noumsi and Etoundi, Christophe Lessouga and Etoundi, Christophe Lessouga and Njitacke, Zeric Tabekoueng and Ele, Pierre},
title = {Frequency- and bit-rate-adaptive compression of electromyography signals by wavelet packet transform and optimal scalar quantizer bit rate allocation},
journal = {Artificial Intelligence in Health},
year = {2026},
volume = {0},
number = {0},
pages = {026140011},
doi = {10.36922/nsce026140011}
}
EUVIMED