PPG Signal Measurement And Analysis By Smartphone

Measuring physiological indices through the smart phone, people reduce their med-ical cost and are provided with a means of routine health examination. In this way, theyprevent the happening of diseases, which alleviate the shortage and uneven distributionof medical resources in our country.Plethysmography (PPG) is a noninvasive measurement of microvascular blood vol-ume change, which can be used to obtain a variety of important physiological indicators,like heart rate, breath rate and heart rate variability etc. Currently, measuring PPG signalthrough the camera on the smart phone is a research hotspot, which not only takes ad-vantage of the sensor and powerful processing capability on the phone, allowing peopleto measure the change of pulse without purchasing additional hardware, but also takesadvantage of the popularity of smart phone, making it possible for people to monitor theirheart rate anytime and anywhere.In this thesis, we mainly discuss the use of smart phone camera to measure PPGsignal in two forms, one is based on the transmission mode, such as measuring fingertipPPG signal, and the otheris based on the reflection mode, measuring PPGsignal remotelyfrom facial video in ambient light. There are mainly two types of noise existed in themeasurement,motionartifactandambientnoise. Thisthesiswilldiscusshowtoovercomethese these noise.The main contribution of this paper is as follows:1. For PPG from fingertip, we propose a feature based on the waveform of the PPGsignal–inverted triangular area (ITA), and we design an adaptive peak detectionalgorithm based on this feature.2. For PPG from face, we use a Kalman filter with a particular structure to eliminateambient noise as well as to enhance the PPG signal.3. We implement these two algorithms on Android platform and developed two appli-cation: Kiwi Cardiac Function and Kiwi Face Cardiac Function Detection.Our experiments show that for PPG signal from fingertip, the proposed method is7.7%higher than the IMS algorithm in sensitivity, and8.3%higher than the ADT algo-rithm in positive predictive value. While for PPG signal from face, the proposed methodis better than the state-of-the-art ICA algorithm in real-time performance and accuracy, which is more suitable for the smart phone application.