Biomedical Signal Processing: Innovation and Applications

✍ Scribed by Iyad Obeid (editor), Ivan Selesnick (editor), Joseph Picone (editor)

• Publisher: Springer
• Year: 2021
• Tongue: English
• Leaves: 265
• Edition: 1st ed. 2021
• Category: Library

✦ Synopsis

This book provides an interdisciplinary look at emerging trends in signal processing and biomedicine found at the intersection of healthcare, engineering, and computer science. It examines the vital role signal processing plays in enabling a new generation of technology based on big data, and looks at applications ranging from medical electronics to data mining of electronic medical records. Topics covered include analysis of medical images, machine learning, biomedical nanosensors, wireless technologies, and instrumentation and electrical stimulation. Biomedical Signal Processing: Innovation and Applications presents tutorials and examples of successful applications, and will appeal to a wide range of professionals, researchers, and students interested in applications of signal processing, medicine, and biology.

✦ Table of Contents

Preface
Contents
1 Multi-class fNIRS Classification of Motor Execution Tasks with Application to Brain-Computer Interfaces
1.1 Introduction
1.1.1 fNIRS
1.1.2 fNIRS-Based BCIs
1.1.2.1 Data Acquisition
1.1.2.2 Data Analysis
1.1.3 Objective
1.2 Experiments
1.2.1 Participants
1.2.2 fNIRS Recording
1.2.3 Experimental Protocol
1.3 Methods
1.3.1 Pre-processing
1.3.2 Feature Extraction
1.3.3 Classification
1.4 Results and Discussions
1.5 Conclusions
References
2 A Comparative Study of End-To-End Discriminative DeepLearning Models for Knee Joint Kinematic Time SeriesClassification
2.1 Introduction
2.2 Related Work
2.2.1 Kinematic Data Pre-processing
2.2.2 Classification
2.3 Methods and Materials
2.3.1 Data Collection
2.3.2 Kinematic Data Pre-processing
2.3.3 Classification
2.3.4 Weighting Imbalanced Classes
2.3.5 Cross-Validation
2.3.6 Performance Measures
2.4 Results
2.4.1 Kinematic Data Pre-processing
2.4.2 Classification
2.5 Discussion and Conclusion
References
3 Nonlinear Smoothing of Core Body Temperature Data with Random Gaps and Outliers (DRAGO)
3.1 Introduction
3.1.1 Measuring Core Body Temperature
3.1.2 Analysis of Core Body Temperature Measurements
3.1.3 Contribution
3.2 Preliminaries
3.2.1 Notation
3.2.2 Encoding Random Gaps in the Input Signal
3.2.3 Majorization-Minimization
3.3 Nonlinear Smoothing of Data with Random Gaps and Outliers (DRAGO)
3.3.1 Problem Formulation
3.3.2 Algorithm
3.3.3 Simulated Example
3.3.4 Parameter Selection
3.4 Estimating Circadian Rhythm Using DRAGO Processed Core Body Temperature Signal
3.5 Conclusion and Future Work
References
4 Wearable Smart Garment Devices for Passive BiomedicalMonitoring
4.1 Introduction
4.1.1 System Deployment
4.2 Functional Fabrics
4.2.1 First Generation Knitted Antenna: Initial Prototype
4.2.1.1 Fabric Material Selection
4.2.1.2 RFID Chip Selection
4.2.1.3 Antenna Design
4.2.2 Second Generation Knitted Antenna: Improved Design 7815769
4.2.2.1 Characteristics of Improved Bellyband
4.2.3 Antenna Characteristics
4.2.4 Radiation Pattern
4.2.5 Fabrication and Sheet Resistance Extraction tajinAWPL
4.2.6 Mitigating On-Body Effects and Signal Degradation
4.2.7 Effect of Sweat and Moisture on Antenna Performance
4.3 A Software Framework for Signal Collection and Processing in the Internet-of-Things
4.3.1 A Sensor Data Framework for Heterogeneous Sensor Communications in the Internet-of-Things
4.3.1.1 Interrogator Drivers
4.3.1.2 Modular Database
4.3.2 A Signal Processing, Multisensor Fusion and Visualization Framework
4.3.2.1 Visualization
4.3.2.2 Detector Processing Module
4.3.2.3 Sensor Fusion and Experimental Protocols with Semi-synthetic Data
4.4 Signal Processing for Sensing and Actuation
4.4.1 RF Signal Model for Biomedical Monitoring
4.4.2 Signal Filtering and Denoising
4.4.2.1 Filtering with an Adaptively Parameterized Savitzky-Golay Filter
4.4.2.2 Denoising with a Reference Tag
4.4.3 Biomedical Applications
4.4.3.1 Activity Classification and Apnea Detection
4.4.3.2 Respiratory Rate Estimation
4.5 Experimental Setup, Results, and Discussion
4.5.1 Experimental Setup
4.5.2 Results and Discussion
4.5.2.1 Semi-unsupervised Classification
4.5.2.2 Respiratory Rate Estimation
4.5.2.3 Square-Wave Generation for Artifact Prediction
4.6 Conclusion and Future Work
References
5 Spatial Distribution of Seismocardiographic Signals
5.1 Introduction
5.2 Methods
5.2.1 Accelerometer Calibration
5.2.2 Experimental Measurements
5.2.3 Preprocessing
5.2.3.1 Filtering
5.2.3.2 SCG Segmentation
5.2.4 Reducing SCG Variability Using Unsupervised Machine Learning
5.2.5 SCG Features
5.2.5.1 SCG Peak-to-Peak Amplitude
5.2.5.2 SCG Signal-to-Noise ratio
5.2.5.3 SCG Morphological Variability
5.2.5.4 Cardiac Timing Intervals (CTIs)
5.2.5.5 SCG 3 Amplitude
5.2.5.6 Maximum Instantaneous Frequency Around SCG 1 and SCG 2 Peak
5.3 Results
5.3.1 SCG Peak-to-Peak Amplitude
5.3.2 Signal-to-Noise Ratio
5.3.3 SCG Morphological Variability
5.3.4 Cardiac Timing Intervals
5.3.5 SCG 3 Amplitude Variation Over Sensor Location
5.3.6 Maximum Instantaneous Frequency Around SCG 1 and SCG 2 Peak
5.3.7 Surface Acceleration Map at Feature Points
5.4 Discussion
5.5 Conclusion
References
6 Determination of Vascular Access Stenosis Location and Severity by Multi-domain Analysis of Blood Sounds
6.1 Introduction and Background
6.2 Prior Work in Phonoangiograhic Detection of Stenosis
6.2.1 Classification of Degree of Stenosis from Phonoangiograms
6.2.2 Localization of Vascular Stenosis from Phonoangiograms
6.3 In Vitro Reproduction of Vascular Bruits
6.4 Signal Processing: Considerations in the Transduction of Bruits
6.4.1 Skin-Coupled Recording Microphone Array
6.4.2 Transducer Front-End Interface Amplifier Design
6.5 Signal Processing and Feature Classification Strategies for Acoustic Detection of Vascular Stenosis
6.5.1 Multi-domain Phonoangiogram Feature Calculation
6.5.1.1 Spectral Domain Feature Extraction
6.5.1.2 Temporospectral Domain Feature Extraction
6.5.1.3 Spatial Domain Feature Extraction
6.6 Classification of Vascular Access Stenosis Location and Severity In Vitro
6.6.1 Multi-domain Feature Selection
6.6.2 Stenosis Spatial Localization Using Acoustic Features
6.6.3 Stenosis Severity Classification from Acoustic Features
6.6.4 Degree of Stenosis Estimation from Acoustic Features
6.7 Conclusion
References
7 Fast Automatic Artifact Annotator for EEG Signals Using Deep Learning
7.1 Introduction
7.2 Related Works
7.3 Method
7.3.1 Resources
7.3.2 Data Preprocessing
7.3.3 Preliminary Studies
7.3.4 Version 1: Recurrent Neural Network Approach
7.3.5 Version 2: Convolutional Neural Network Approach
7.3.6 Ensemble Method
7.4 Results and Discussion
7.4.1 Recurrent Neural Network-Based Classifier
7.4.2 Convolutional Neural Network-Based Classifier
7.4.3 Ensemble Method
7.5 Conclusion
References
8 Objective Evaluation Metrics for Automatic Classification of EEG Events
8.1 Introduction
8.2 Basic Error Measures and Relevant Derived Measures
8.3 Evaluation Metrics
8.3.1 NIST Actual Term-Weighted Value (ATWV)
8.3.2 Dynamic Programming Alignment (DPALIGN)
8.3.3 Epoch-Based Sampling (EPOCH)
8.3.4 Any-Overlap Method (OVLP)
8.3.5 Time-Aligned Event Scoring (TAES)
8.3.6 Interrater Agreement (IRA)
8.3.7 A Brief Comparison of Metrics
8.4 Evaluation
8.4.1 The TUH EEG Seizure Corpus
8.4.2 Machine Learning Architectures
8.5 Derived Measures
8.5.1 Detection Error Trade-off Analysis
8.5.2 Accuracy and Other Derived Scores
8.5.3 Additional Insight
8.6 Statistical Analysis
8.6.1 Kolmogorov-Smirnov and Pearson's R Tests
8.6.2 Z-Tests
8.7 Conclusions
References
Index