Heartbeat detection in multimodal physiological signals using signal quality assessment based on sample entropy

Area of Science:

• Biomedical Engineering
• Physiological Signal Processing
• Cardiovascular Monitoring

Background:

• Accurate heartbeat detection is crucial for estimating cardiovascular parameters.
• Traditional R-peak detection from electrocardiogram (ECG) signals can be challenging due to artifacts.
• Existing methods often process multimodal physiological signals independently, neglecting valuable correlated information.

Purpose of the Study:

• To develop a novel technique for enhanced heartbeat identification using multimodal data.
• To improve the accuracy and robustness of R-peak detection by fusing ECG and arterial blood pressure (ABP) signals.
• To leverage sensor-fusion to overcome limitations of unimodal signal processing in noisy conditions.

Main Methods:

• A sensor-fusion technique combining R-peaks from ECG and systolic peaks from ABP signals was developed.
• Sample entropy (SampEn) was employed as a metric to assess signal noise content.
• The proposed method was evaluated on 100 records from the Computing in Cardiology Challenge 2014 dataset.

Main Results:

• The multimodal beat detector achieved high performance: gross sensitivity (Se) of 99.65% and positive predictivity (PPV) of 99.91%.
• These results demonstrate superior or comparable performance to unimodal detectors, particularly in challenging signal conditions.
• The fusion approach effectively enhanced the efficiency and accuracy of heartbeat detection.

Conclusions:

• Multimodal signal fusion offers a significant advantage for robust heartbeat detection compared to unimodal approaches.
• The proposed method provides a reliable solution for accurate R-peak detection in clinical settings with potential for artifact reduction.
• This technique can improve the estimation of heart rate and heart rate variability from combined ECG and ABP signals.