Accurate Fiducial Point Detection Using Haar Wavelet for Beat-by-Beat Blood Pressure Estimation

Area of Science:

• Biomedical Engineering
• Physiological Monitoring
• Signal Processing

Background:

• Cuff-less blood pressure (BP) measurement using Pulse Arrival Time (PAT) from Electrocardiogram (ECG) and Photoplethysmogram (PPG) is a widely accepted technique.
• Conventional methods extract features from isolated pulses, resulting in intermittent BP estimations.
• The reliance on PAT alone can lead to inaccurate Systolic Blood Pressure (SBP) and Diastolic Blood Pressure (DBP) estimations.

Purpose of the Study:

• To develop a method for uninterrupted BP measurements by extracting features from each beat of ECG and PPG signals.
• To improve fiducial point detection precision and attenuate noise using Haar transformation.
• To propose supplementary features highly correlated with physiological parameters for accurate BP estimation.

Main Methods:

• Feature extraction from individual ECG and PPG beats using an Adaptive Window Wavelet Transformation (AWWT) based on Haar wavelet transformation.
• Segregation of pulses and noise attenuation for enhanced signal quality.
• Development of a log-linear regression algorithm to process extracted features for beat-to-beat BP calculation.
• Data collection from 171 diverse human subjects adhering to Association for the Advancement of Medical Instrumentation (AAMI) guidelines.

Main Results:

• Achieved mean errors of 0.43 mmHg for SBP and 0.20 mmHg for DBP.
• Attained mean absolute errors of 4.6 mmHg for SBP and 2.3 mmHg for DBP.
• Reported standard deviations of 6.13 mmHg for SBP and 3.06 mmHg for DBP, indicating high precision.

Conclusions:

• The extracted features are highly precise, and the evaluated BP values meet AAMI standards.
• This continuous, real-time BP monitoring technique offers significant clinical utility.
• The method is particularly beneficial for managing hypertensive and pre-hypertensive individuals.