An Algorithm for Real-Time Pulse Waveform Segmentation and Artifact Detection in Photoplethysmograms

Area of Science:

• Biomedical Engineering
• Signal Processing
• Medical Device Technology

Background:

• Photoplethysmography (PPG) is vital for medical monitoring, but artifacts often compromise signal integrity.
• Existing artifact detection algorithms have limitations in design and scope.
• Reliable artifact identification is crucial for accurate PPG-based medical measurements.

Purpose of the Study:

• To develop and validate a novel embedded algorithm for real-time pulse waveform (PWF) segmentation and artifact detection.
• To address limitations in current PPG artifact detection methods.
• To improve the reliability and utility of PPG signals in medical applications.

Main Methods:

• Developed a time-domain contour analysis algorithm for PWF segmentation and artifact detection.
• Implemented the algorithm on a 32-bit ARM core microcontroller for real-time processing.
• Validated the algorithm using PPG data from 63 subjects across diverse clinical settings (sleep, ergometry, ICU).

Main Results:

• The algorithm achieved high performance in beat-to-beat comparisons: 99.6% sensitivity, 90.5% specificity, 98.5% precision, and 98.3% accuracy.
• Excellent interrater agreement was demonstrated with Cohen's kappa of 0.927 and F-measure of 0.990.
• The algorithm effectively segmented PWF and detected artifacts, outperforming existing methods.

Conclusions:

• The developed PWF analysis is a robust method for determining PPG signal quality.
• The algorithm supports real-time annotation, data compression, and calculation of pulse wave metrics.
• This approach enhances the diagnostic value and applicability of PPG in various medical fields.