PPGTempStitch: A MATLAB Toolbox for Augmenting Annotated Photoplethsmogram Signals.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Medical Informatics

Background:

• Accurate annotation of photoplethysmogram (PPG) waveforms, specifically identifying onset and systolic peaks, is crucial for evaluating PPG algorithm performance.
• A significant limitation in PPG research is the scarcity of publicly available datasets with reliably annotated waveforms.

Purpose of the Study:

• To develop a novel MATLAB toolbox for synthesizing long, annotated PPG signals by stitching existing annotated PPG templates.
• To enhance the diversity and size of annotated PPG datasets for improved algorithm development and validation.

Main Methods:

• The toolbox utilizes four types of annotated PPG templates (regular, irregular, fast rhythm, noisy) and stitches them randomly.
• It simulates realistic PPG signals by incorporating variable noise levels and waveform variations.
• Two stitching methods (systolic peak-based and onset-based) are implemented, with cubic spline interpolation for smoothing and a skewness index for quality control.

Main Results:

• The developed toolbox successfully generates long, annotated PPG signals from combinations of predefined templates.
• It allows for the simulation of diverse PPG signal characteristics, including different rhythms and noise levels.
• The open-source toolbox with a graphical user interface provides a practical solution for data augmentation in PPG research.

Conclusions:

• The stitching-based synthesis method is an effective data augmentation strategy for PPG signals.
• This approach significantly expands the availability of diverse, annotated PPG data for training and testing feature extraction algorithms.
• The free and open-source toolbox facilitates advancements in PPG signal analysis and algorithm development.