A procedure for denoising dual-axis swallowing accelerometry signals

Area of Science:

• Biomedical Engineering
• Signal Processing
• Medical Device Technology

Background:

• Dual-axis swallowing accelerometry is a developing technology for assessing dysphagia.
• Swallowing signals often contain noise from physiological and motion artifacts, hindering accurate analysis.

Purpose of the Study:

• To propose a novel, computationally efficient denoising scheme for dual-axis swallowing accelerometry signals.
• To improve the accuracy and reliability of dysphagia assessment using accelerometry data.

Main Methods:

• A novel denoising scheme using a computationally efficient search for optimal denoising thresholds within a reduced wavelet subspace.
• Algorithm determines subspace viability using the minimum value of the estimated upper bound for reconstruction error.

Main Results:

• The proposed scheme is more computationally efficient than Minimum Noiseless Description Length (MNDL)-based denoising.
• It yields smaller reconstruction errors compared to MNDL, SURE, and Donoho denoising methods.
• Improved performance was observed for dry, wet, and wet chin tuck swallows using real accelerometry data.

Conclusions:

• The developed denoising scheme offers superior computational efficiency and accuracy for swallowing accelerometry signals.
• This advancement is crucial for the future development of medical devices for dysphagia assessment.