PPG Signal Analysis: An Introduction Using MATLAB

Cover
Half Title
Title Page
Copyright Page
Dedication
Table of Contents
List of Figures and Tables
Preface
Acknowledgments
The Author
How to Use This Book?
Chapter 1: Math Foundations
	1.1 Learning Objectives
	1.2 Scalars
		1.2.1 Scalar Mathematical Operations
		1.2.2 Assigning Scalar Values
	1.3 Vectors
		1.3.1 Vector Mathematical Operations
		1.3.2 Assigning Vector Elements
		1.3.3 Assigning Vector Elements Using a Function
		1.3.4 Assigning Vector Elements Using a Colon (:)
		1.3.5 Addressing Vector Elements
		1.3.6 Increasing the Vector Size
	1.4 Matrices
		1.4.1 Matrix Mathematical Operations
		1.4.2 Assigning Matrix Elements
		1.4.3 Assigning Matrix Elements Using a Function
		1.4.4 Addressing Matrix Elements
	1.5 Relational Operators
	1.6 NaN
	1.7 Strings
	1.8 Structures
	1.9 Cell
	1.10 Import/Export Data
	1.11 Workspace User Input
Chapter 2: Photoplethysmogram Signals
	2.1 Learning Objectives
	2.2 Background
	2.3 Oxygen Transport
	2.4 Terminologies and Acronyms
		2.4.1 DVP
		2.4.2 PTG
		2.4.3 SDPTG
		2.4.4 APG
		2.4.5 SDDVP
		2.4.6 Terminology Selection and Search Strategy
		2.4.7 Standard Acronyms
	2.5 Why PPG Signal?
	2.6 Plethysmography Types
	2.7 Measuring Sites
	2.8 Modes of PPG Measurement
		2.8.1 Transmissive Mode
		2.8.2 Reflective Mode
	2.9 Calculation of Oxygen Saturation
	2.10 Simulation of PPG Signal Using Sinusoids
	2.11 Simulation of PPG Signal using Two Gaussian Functions
	2.12 PPG Sensors
		2.12.1 Probe-Based PPG Signals
		2.12.2 Video-Based PPG Signals
	2.13 Current Challenges
		2.13.1 Powerline Interference
		2.13.2 Sudden Amplitude Change
		2.13.3 Motion Artifact
		2.13.4 Multi-Parameter Systems
		2.13.5 Research Design
	2.14 Summary
Chapter 3: Visualization of PPG Signals
	3.1 Learning Objectives
	3.2 Plot
	3.3 Bar
	3.4 Area
		3.4.1 Histogram
	3.5 Periodogram
	3.6 Spectrogram
		3.6.1 Wavelets
	3.7 Eventogram
	3.8 Discussion
	3.9 Summary
Chapter 4: Pre-processing of PPG Signals
	4.1 Learning Objectives
	4.2 Filter Types
		4.2.1 Moving Average (MA) Filter
		4.2.2 Butterworth Filter (Butter)
		4.2.3 Chebyshev Filter (Cheby I and Cheby II)
		4.2.4 Elliptic Filter (Ellip)
		4.2.5 General Comment
	4.3 Filter Design
		4.3.1 Low-Pass Filter
		4.3.2 High-Pass Filter
		4.3.3 Band-Pass Filter
		4.3.4 Band-Stop Filter
	4.4 Convolution
		4.4.1 Improving PPG Beat Quality
		4.4.2 Filtering PPG Signal
	4.5 Cross-correlation
		4.5.1 Filtering One PPG Beat
		4.5.2 Filtering PPG Signal Quality
	4.6 Summary
Chapter 5: Signal Quality Assessment
	5.1 Learning Objectives
	5.2 Introduction
	5.3 Annotation
	5.4 Signal Quality Indices
		5.4.1 Perfusion ( P SQI)
		5.4.2 Skewness (S SQI):
		5.4.3 Kurtosis (K SQI)
		5.4.4 Entropy (E SQI)
		5.4.5 Zero Crossing Rate (Z SQI)
		5.4.6 Signal-to-Noise Ratio (N SQI)
		5.4.7 Matching Systolic Detectors (M SQI)
		5.4.8 Relative Power (R SQI)
	5.5 Summary
Chapter 6: PPG Feature Extraction
	6.1 Learning Objectives
	6.2 Overview of PPG Features
	6.3 Features of PPG Waveforms
		6.3.1 Systolic Amplitude
		6.3.2 Pulse Width
		6.3.3 Pulse Area
		6.3.4 Peak-to-Peak Interval
		6.3.5 Pulse Interval
		6.3.6 Augmentation Index
		6.3.7 Large Artery Stiffness Index
	6.4 Features of VPG Signals
		6.4.1 Diastolic Point
		6.4.2 Δ T Calculation
		6.4.3 Crest Time Calculation
	6.5 Features of APG Signals
		6.5.1 a, b, c, d, and e Waves
		6.5.2 Ratio b / a Index
		6.5.3 Ratio c / a Index
		6.5.4 Ratio d / a Index
		6.5.5 Ratio e / a Index
		6.5.6 Ratio ( b  −  c  −  d  −  e)/ a Index
		6.5.7 Ratio ( b  −  e)/ a Index
		6.5.8 Ratio ( b  −  c  −  d)/ a Index
		6.5.9 Ratio ( c  +  d  −  b)/ a Index
		6.5.10 aa Interval
		6.5.11 APG Beat Waveform
		6.5.12 Segment of APG Signal
		6.5.13 Chaos Attractor
		6.5.14 MATLAB Functions for Features Extraction
		6.5.15 MATLAB Code for Extracting 125 PPG Features
			6.5.15.1 Time Span
			6.5.15.2 Features of PPG Amplitude
			6.5.15.3 Features of VPG and APG
			6.5.15.4 Waveform Area
			6.5.15.5 Power Area
			6.5.15.6 Ratio
			6.5.15.7 Slope
			6.5.15.8 Code for PPG Feature Calculation
			6.5.15.9 Heart Rate Variability
			6.5.15.10 Time Domain Methods
			6.5.15.11 Frequency Domain Methods
		6.5.16 Nonlinear Methods
			6.5.16.1 Poincaré Plot
			6.5.16.2 Approximate Entropy and Sample Entropy
		6.5.17 Discussion
	6.6 Summary
Chapter 7: A Generic Method for Event Detection
	7.1 Learning Objectives
	7.2 Introduction
	7.3 Data Used
	7.4 TERMA Framework
		7.4.1 Prior Knowledge
		7.4.2 Bandpass Filter
		7.4.3 Signal Enhancement
		7.4.4 Generating Blocks of Interest
		7.4.5 Thresholding
		7.4.6 Detecting Event Peak
	7.5 Results
		7.5.1 Training Results
		7.5.2 Testing
	7.6 Discussion
		7.6.1 Frequency Band Choice
		7.6.2 Window Size Choice
		7.6.3 Offset β Choice
		7.6.4 Battery-Driven Devices
		7.6.5 Optimization Step
			7.6.5.1 Exhaustive Search
			7.6.5.2 Gradient-Based Search
			7.6.5.3 Parallel Execution
	7.7 Significance of TERMA
	7.8 Summary
Chapter 8: Feature Selection
	8.1 Learning Objectives
	8.2 Feature Normalization
		8.2.1 Linear Normalization
		8.2.2 Nonlinear Normalization
	8.3 Criteria for Selection and Evaluation
		8.3.1 Independent Student’s t -test
		8.3.2 Dependent Samples (Paired) t -test
		8.3.3 Receiver Operating Characteristic Curve
		8.3.4 Analysis of Variance (ANOVA)
		8.3.5 Fisher’s Measure
		8.3.6 Divergence Measure
		8.3.7 Bhattacharyya’s Measure
		8.3.8 Scatter Measure
	8.4 Optimal Feature(s)
		8.4.1 Individual Feature Selection
	8.5 Search Method
		8.5.1 Optimal Search
		8.5.2 Suboptimal Search
	8.6 Summary
Chapter 9: Identifying Adverse Events
	9.1 Learning Objectives
	9.2 Minimum Distance Classifier
	9.3 Bayes Classifier
	9.4 Competitive Neural Network
	9.5 Discriminant Analysis
	9.6 Other Classifiers
	9.7 Classification Example using Classical Machine Learning Methods
	9.8 Classification Example using Deep Learning
	9.9 Effectiveness Evaluation
		9.9.1 K-Fold Cross Validation
		9.9.2 Class Imbalance
		9.9.3 Confusion Matrix
		9.9.4 Sensitivity versus Specificity
	9.10 Summary
Chapter 10: Application of PPG to Global Health
	10.1 Learning Objectives
	10.2 Introduction
	10.3 Overview
	10.4 Simplicity
	10.5 Mining
	10.6 Connection
	10.7 Reliability
	10.8 Affordability
	10.9 Scalability
	10.10 Noncommunicable Disease Case Studies
		10.10.1 Case I: Detection of Heat Stress in a Changing Climate
			10.10.1.1 Simplicity
			10.10.1.2 Mining
			10.10.1.3 Connection
			10.10.1.4 Reliability
			10.10.1.5 Affordability
			10.10.1.6 Scalability
		10.10.2 Case II: Prediction of Adverse Outcomes Related to Preeclampsia using SpO2
			10.10.2.1 Simplicity
			10.10.2.2 Mining
			10.10.2.3 Connection
			10.10.2.4 Affordability
			10.10.2.5 Scalability
		10.10.3 Case III: Hypertension Risk Stratification
			10.10.3.1 Simplicity
			10.10.3.2 Mining
			10.10.3.3 Connection
			10.10.3.4 Affordability
			10.10.3.5 Scalability
	10.11 User Performance
	10.12 Summary
Chapter 11: Available PPG Databases
	11.1 Fingertip PPG from Hypertensive Subjects
	11.2 Fingertip PPG from an Intensive Care Unit
	11.3 Wrist PPG During Exercise
	11.4 Fingertip PPG and Respiration
		11.4.1 The University of Queensland Vital Signs Dataset
		11.4.2 BioSec.Lab PPG Dataset
		11.4.3 Vortal Dataset
	11.5 Summary
References
Index