ppg-signal-extraction

Extraction of photoplethysmogram (PPG) signal from camera recording and assessment of PPG signal quality

Figure 1: Example of extracted PPG signal (the good part of the signal is colored in green)

Introduction

To estimate the vital signs (e.g. blood pressure or blood glucose level) based on photoplethysmogram (PPG) signal from camera recording, we can built a data-driven model. This repository contains some scripts to create a dataset of measurements on which the model can be trained.

Description

To capture PPG signal from a camera, you can try using the Python or Bash scripts in scripts directory:

• capture-noir.py - extracts frames from video captured by Raspberry Pi device using the camera without infrared filter;
• convert-iphone.sh - uses ffmpeg to extract frames from video captured by iPhone camera (video in QTFF format);
• save-measurement.py - extracts the PPG signal from provided video frames (images in PNG format) and saves the signal to a CSV file;

For more details, see the notebooks (written in R):

• Definitions.ipynb - example of generated dataset, PPG signal and the assessment of signal quality shown in Figure 1;
• Analysis.ipynb - more examples of extracted PPG signals;