Please use this identifier to cite or link to this item: https://hdl.handle.net/10356/76376
Title: PPG signal classification for motion artefact detection
Authors: Li, Longjie
Keywords: DRNTU::Engineering::Electrical and electronic engineering
Issue Date: 2018
Abstract: Photoplethysmography (PPG) signal is usually obtained by using a light source to illuminate the skin. PPG is a noninvasive technique, it has received more and more attention in recent years because it can give some basic but important cardiovascular parameters such as heart rate, blood pressure, oxygen saturation, cardiac output and so on. PPG is widely used in smart phone and wearable devices for health monitoring due to its simplicity and low cost. However, PPG suffers from motion artefact thus a motion artefact detection system is required. In this project, a PPG signal classification system for motion artefact detection is proposed. The PPG signal is first modeled as a periodic signal with fundamental frequency and four harmonics, then a DTFT based algorithm for Fourier coefficients (amplitude and phase) estimation is discussed and validated using the PPG signal collected in laboratory. The use of the model and algorithm allows the morphological information of PPG signal to be correctly extracted avoiding complex interpolation and decimation. The SVM is used as classifier training algorithm due to its simplicity. A frequency domain normalization which is achieved by forcing the amplitude and phase of the fundamental frequency to be 1 and 0 is applied before computing z-score. Experiments show that the use of two-step normalization (frequency domain followed by z-score) can enhance the accuracy. All the samples used in this project (1363 samples from 26 subjects) are manually labeled according to its morphology. The proposed classification system can distinguish the normal PPG signal from motion artefacts with 96.55% accuracy.
URI: http://hdl.handle.net/10356/76376
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:EEE Theses

Files in This Item:
File Description SizeFormat 
LiLongjie_2018.pdf
  Restricted Access
Main article2.41 MBAdobe PDFView/Open

Page view(s) 1

79
checked on Oct 22, 2020

Download(s) 1

12
checked on Oct 22, 2020

Google ScholarTM

Check

Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.