A strong anti-noise segmentation algorithm based on variational mode decomposition and multi-wavelet for wearable...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Wearable Technology

Background:

• Wearable heart sound acquisition systems (WHSAS) are increasingly used but susceptible to significant noise.
• Common noise sources include Gaussian white noise, powerline interference, colored noise, motion artifacts, and lung sounds.
• High-precision segmentation is crucial for WHSAS in noisy environments, especially during physical activity.

Purpose of the Study:

• To develop and validate a high-precision heart sound segmentation algorithm for WHSAS operating in noisy conditions.
• To address the challenges posed by mixed noise sources encountered during physical activities.

Main Methods:

• A novel segmentation algorithm combining Variational Mode Decomposition (VMD) and multi-wavelet analysis.
• VMD is used for layered noise filtering.
• Multi-wavelet analysis constructs a time-frequency matrix, followed by principal component analysis for dimensionality reduction.
• Extraction of high-order Shannon envelope and Teager energy envelope for signal segmentation.

Main Results:

• The proposed algorithm successfully filters various noise types using VMD.
• Principal component analysis effectively reduces the dimensionality of the time-frequency matrix.
• Accurate heart sound segmentation was achieved under mixed noise conditions using the developed WHSAS.
• The algorithm demonstrated high precision in segmenting heart sound signals.

Conclusions:

• The VMD and multi-wavelet-based algorithm provides a robust solution for precise heart sound segmentation in WHSAS.
• This method is effective in mitigating noise interference common in wearable health monitoring during physical activities.
• The findings support the advancement of wearable technology for accurate cardiac auscultation.