How to Evaluate Signal Quality of ear-ECG?

Area of Science:

• Biomedical Engineering
• Signal Processing
• Wearable Technology

Background:

• Hearable devices offer convenient heart rate (HR) estimation via ear-based electrocardiogram (ECG).
• Ear-ECG signals suffer from low amplitude and interference, impacting accuracy compared to traditional ECG.
• Existing signal quality assessment methods require adaptation for the unique characteristics of ear-ECG.

Purpose of the Study:

• To comprehensively evaluate ear-ECG signal quality using adapted metrics.
• To develop a simplified, practical approach for assessing ear-ECG signal quality.
• To validate the performance of a deep-matched filter for ear-ECG R-peak detection.

Main Methods:

• Adapted 13 signal quality metrics from standard ECG, including higher-order statistics, frequency-based, and template-based indexes.
• Utilized a deep-matched filter detector specifically designed for ear-ECG R-peak detection.
• Collected 5-minute resting ear-ECG and standard ECG data from 40 healthy subjects (<35 years).
• Categorized signal segments based on R-peak detection sensitivity relative to a reference ECG.

Main Results:

• Proposed a simplified four-metric approach for ear-ECG quality assessment.
• Key metrics include deep-matched filter output, QRS band power ratio, ECG band power ratio, and baseline drift power ratio.
• The proposed method demonstrated a lower error rate and balanced computational complexity with robust performance.

Conclusions:

• A practical and accurate method for assessing ear-ECG signal quality has been developed.
• The simplified approach, utilizing deep-matched filters, enhances the reliability of HR estimation from hearable devices.
• This method offers a balanced solution combining low complexity, interpretability, and robust performance for ear-ECG analysis.