Global-key region collaborative noise-aware for electrocardiogram signal quality assessment

Area of Science:

• Biomedical Engineering
• Signal Processing
• Artificial Intelligence

Background:

• Wearable electrocardiogram (ECG) devices offer continuous cardiac monitoring but are susceptible to noise, impacting diagnostic accuracy.
• Current supervised ECG signal quality assessment (SQA) methods require extensive labeled data, which is impractical for real-world applications.
• Existing unsupervised methods often overlook critical noise-specific regions, limiting their ability to differentiate noise from physiological signals.

Purpose of the Study:

• To develop an unsupervised ECG signal quality assessment (SQA) method that overcomes the limitations of existing approaches.
• To improve the reliability of automated cardiac diagnosis from noisy wearable ECG data.
• To enhance the generalization capability of SQA models to unseen, low-quality ECG signals.

Main Methods:

• Proposed a global-key region collaborative noise-aware ECG signal quality assessment (NA-SQA) method for unsupervised settings.
• Utilized an autoencoder to reconstruct high-quality ECG signals from noisy inputs, generating artificial noise for training.
• Integrated global signal information and discriminative noise-region features for joint quality prediction.

Main Results:

• The NA-SQA method achieved high F1 scores across four public datasets (BUTQDB, Icentia11K, EHOQA, EAWQA), reaching up to 91.42%.
• Demonstrated consistent outperformance compared to state-of-the-art unsupervised ECG quality assessment techniques.
• Showcased superior robustness and generalization capabilities on unseen noisy ECG signals.

Conclusions:

• The proposed NA-SQA method effectively integrates global and key noise-region features for accurate unsupervised ECG signal quality assessment.
• This approach significantly enhances the reliability of automated diagnosis from wearable ECG devices in noisy environments.
• NA-SQA offers a robust and generalizable solution for real-world ECG monitoring applications.