Biomedical Signal Processing and Artificial Intelligence in Healthcare

Walid A. Zgallai

• Publisher: Academic Press
• Year: 2020
• Print ISBN: 9780128189467
• eBook ISBN: 9780128189474
• Topic: Tecnologia e ingegneria
• Subtopic: Ingegneria generale

Biomedical Signal Processing and Artificial Intelligence in Healthcare is a new volume in the Developments in Biomedical Engineering and Bioelectronics series. This volume covers the basics of biomedical signal processing and artificial intelligence. It explains the role of machine learning in relation to processing biomedical signals and the applications in medicine and healthcare. The book provides background to statistical analysis in biomedical systems. Several types of biomedical signals are introduced and analyzed, including ECG and EEG signals. The role of Deep Learning, Neural Networks, and the implications of the expansion of artificial intelligence is covered. Biomedical Images are also introduced and processed, including segmentation, classification, and detection. This book covers different aspects of signals, from the use of hardware and software, and making use of artificial intelligence in problem solving.Dr Zgallai's book has up to date coverage where readers can find the latest information, easily explained, with clear examples and illustrations. The book includes examples on the application of signal and image processing employing artificial intelligence to Alzheimer, Parkinson, ADHD, autism, and sleep disorders, as well as ECG and EEG signals. Developments in Biomedical Engineering and Bioelectronics is a 10-volume series which covers recent developments, trends and advances in this field. Edited by leading academics in the field, and taking a multidisciplinary approach, this series is a forum for cutting-edge, contemporary review articles and contributions from key 'up-and-coming' academics across the full subject area. The series serves a wide audience of university faculty, researchers and students, as well as industry practitioners.- Coverage of the subject area and the latest advances and applications in biomedical signal processing and Artificial Intelligence- Contributions by recognized researchers and field leaders- On-line presentations, tutorials, application and algorithm examples

Chapter 1: Introduction to biomedical signal processing and artificial intelligence

Dr. Noura AlHinai Higher Colleges of Technology, Dubai, United Arab Emirates

Abstract

This chapter provides an overall review of biomedical signal processing using artificial intelligence focusing on various organs of the body. The biosignals are analyzed using different assessment methods, such as, electrocardiogram (ECG/EKG) and electroencephalogram (EEG). The signals are small and reach the sensors attenuated and with noise; hence, there is a need for amplifiers that are used to amplify the signals and can be used for human computer interaction. Since the biosignals are weak in level, they are easily distorted by noise. The most common noise types, thermal noise and flicker noise, are discussed further in this chapter. Later in this chapter, mitigation techniques such as finite impulse response (FIR) filters and Butterworth filters are applied to reduce noise in ECG signal. The objective of computer-aided diagnosis (CAD) is addressed to decrease the rate of false diagnosis by assisting physicians with a second opinion.

Several studies revealed the importance of integrating artificial intelligence systems in biomedical signal processing applications and provided insight solutions to minimize the challenges faced by physician when making a diagnosis. The concept of fuzzy logic has common features with neural networks when it comes to mimicking human behavior. Therefore fuzzy logic can be considered a branch of artificial intelligence, especially in situations of vagueness. Reduction of noise using different filtering techniques produce improved readings for disease detection, which assists physicians to better diagnose.

Keywords

Signal processing; Biosignals; ECG; EEG; Thermal noise; Flicker noise; FIR; Butterworth; AI; Fuzzy logic

Abbreviations

1D one-dimensional

AI artificial intelligence

ASICs application-specific integrated circuits

AV atrioventricular

BPM beats per minute

CAD computer-aided diagnosis

DL deep learning

ECG/EKG electrocardiogram

EEG electroencephalogram

EMF electromagnetic field

FIR finite impulse response

Hz hertz

IIR infinite impulse response

m-D multidimensional

ML machine learning

PLI power-line interference

SA sinoatrial

1.1: Introduction to signal processing

A signal is a mathematical function of one or more independent variables, representing a measureable quantity that can propagate in a certain medium [1]. Signals can be classified in many ways based on different parameters, such as time, periodicity of signal, nature of certainty, and causality. Subsequently signals can be classified as continuous time signals, discrete time signals, or digital signals as shown in Fig. 1.1[2].

Fig. 1.1 Signal classification.

A continuous time signal is also known as an analog signal, where time and amplitude are continuous. Hence, time is an independent variable that belongs to real values. A discrete time signal is a signal that has been sampled at discrete intervals of time. Hence, time is discrete, and amplitude is continuous. A digital signal is where both time and amplitude are quantized into discrete signal levels [2].

Signal processing involves manipulating a signal to change the basic characteristics of a signal or to extract some information from it. This is usually done by either using a computer program, application-specific integrated circuits (ASICs), or analog electrical circuit [3]. Software algorithms have advantage over analog electrical circuits in that they can be adapted for different scenarios and situations. The applications of signal processing are almost as diverse as the number of signals there are themselves. In the medical field, signal processing plays an important role in imaging, as well as monitoring, for example, the electrical activity in the heart and in the brain. There are three different classes of typical signal processing problems [4]:

1. (1) Eliminating noise: A noisy electrocardiograph can exhibit discontinuous behavior of the recorded signal. We know from the biology that the electrical activity of the heart should behave in a smooth fashion. Thus the goal of signal processing would be to eliminate or reduce noise and produce a clean signal that reflects the true underlying activity of the heart in a patient.
2. (2) Correction of distortion: Running a blurry image through a signal processing algorithm that can reconstruct a sharper, more focused image. Correcting distortion of images can be obvious; however, this can also be applied to signals being distorted in time.
3. (3) Extracting information embedded within the measured signals: For example, using a radar system to determine an aircraft position and velocity. Firstly, the position of the aircraft is governed by the time delay that it took for a pulse to travel from the radar to the airplane and back, and knowing the speed of light, we can figure out how far away it was. Secondly the relative velocity of the airplane with respect to the radar is embedded in the Doppler shift that can be seen in the received pulse.

In conclusion, signals and signal processing encompass every aspect of our lives. Signal processing is often used to address three different problems: (1) reduce noise in measured signals, (2) correct distortion, and (3) extract inf...

Table of contents

• Cover image
• Title page
• Table of Contents
• Copyright
• Dedication
• Contributors
• Foreword
• Preface
• Chapter 1: Introduction to biomedical signal processing and artificial intelligence
• Chapter 2: Characterization of biomedical signals: Feature engineering and extraction
• Chapter 3: Supervised and unsupervised learning
• Chapter 4: Machine learning in biomedical signal processing with ECG applications
• Chapter 5: Deep EEG: Deep learning in biomedical signal processing with EEG applications
• Chapter 6: Fuzzy logic in medicine
• Chapter 7: Neural network applications in medicine
• Chapter 8: Analysis and management of sleep data
• Index