RANGARAJ M. RANGAYYAN is Professor Emeritus of Electrical and Computer Engineering, University of Calgary. Dr. Rangayyan has developed several algorithms for biomedical signal and image processing for computer-aided diagnosis. He is a Life Fellow of the IEEE, Fellow of the Royal Society of Canada, and Fellow of the Canadian Medical and Biological Engineering Society, and has been recognized with several other fellowships and awards, including the Outstanding Engineer Medal of IEEE Canada. SRIDHAR KRISHNAN is Professor in the Electrical, Computer, and Biomedical Engineering Department, Toronto Metropolitan University, Canada. He served as Associate Dean (Research and Development), Faculty of Engineering and Architectural Science, and is the Founding Co-Director of the Institute for Biomedical Engineering, Science, and Technology. Dr. Krishnan held the Canada Research Chair position in Biomedical Signal Analysis and is a Fellow of the Canadian Academy of Engineering.

• About the Authors xviForeword by Prof. Willis J. Tompkins xviiiForeword by Prof. Alan V. Oppenheim xixPreface xxiiAcknowledgments xxviiiSymbols and Abbreviations xxxiAbout the Companion Website xxxix1 Introduction to Biomedical Signals 11.1 The Nature of Biomedical Signals 11.2 Examples of Biomedical Signals 41.2.1 The action potential of a cardiac myocyte 51.2.2 The action potential of a neuron 91.2.3 The electroneurogram (ENG) 101.2.4 The electromyogram (EMG) 121.2.5 The electrocardiogram (ECG) 201.2.6 The electroencephalogram (EEG) 291.2.7 Event-related potentials (ERPs) 351.2.8 The electrogastrogram (EGG) 361.2.9 The phonocardiogram (PCG) 371.2.10 The carotid pulse 401.2.11 The photoplethysmogram (PPG) 411.2.12 Signals from catheter-tip sensors 431.2.13 The speech signal 441.2.14 The vibroarthrogram (VAG) 481.2.15 The vibromyogram (VMG) 521.2.16 Otoacoustic emission (OAE) signals 521.2.17 Bioacoustic signals 521.3 Objectives of Biomedical Signal Analysis 521.4 Challenges in Biomedical Signal Analysis 551.5 Why Use Computer-aided Monitoring and Diagnosis? 581.6 Remarks 601.7 Study Questions and Problems 601.8 Laboratory Exercises and Projects 62References 632 Analysis of Concurrent, Coupled, and Correlated Processes 712.1 Problem Statement 712.2 Illustration of the Problem with Case Studies 722.2.1 The ECG and the PCG 722.2.2 The PCG and the carotid pulse 732.2.3 The ECG and the atrial electrogram 732.2.4 Cardiorespiratory interaction 752.2.5 Heart-rate variability 752.2.6 The EMG and VMG 772.2.7 The knee-joint and muscle-vibration signals 772.3 Application: Segmentation of the PCG 782.4 Application: Diagnosis and Monitoring of Sleep Apnea 792.4.1 Monitoring of sleep apnea by polysomnography 802.4.2 Home monitoring of sleep apnea 802.4.3 Multivariate and multiorgan analysis 822.5 Remarks 852.6 Study Questions and Problems 852.7 Laboratory Exercises and Projects 86References 863 Filtering for Removal of Artifacts 913.1 Problem Statement 913.2 Random, Structured, and Physiological Noise 923.2.1 Random noise 923.2.2 Structured noise 983.2.3 Physiological interference 983.2.4 Stationary, nonstationary, and cyclostationary processes 993.3 Illustration of the Problem with Case Studies 1013.3.1 Noise in event-related potentials 1023.3.2 High-frequency noise in the ECG 1023.3.3 Motion artifact in the ECG 1023.3.4 Power-line interference in ECG signals 1033.3.5 Maternal ECG interference in fetal ECG 1053.3.6 Muscle-contraction interference in VAG signals 1053.3.7 Potential solutions to the problem 1063.4 Fundamental Concepts of Filtering 1063.4.1 Linear shift-invariant filters and convolution 1073.4.2 Transform-domain analysis of signals and systems 1173.4.3 The pole–zero plot 1233.4.4 The Fourier transform 1253.4.5 The discrete Fourier transform 1263.4.6 Convolution using the DFT 1313.4.7 Properties of the Fourier transform 1333.5 Synchronized Averaging 1353.6 Time-domain Filters 1393.6.1 Moving-average filters 1393.6.2 Derivative-based operators to remove low-frequency artifacts 1453.6.3 Various specifications of a filter 1523.7 Frequency-domain Filters 1533.7.1 Removal of high-frequency noise: Butterworth lowpass filters 1543.7.2 Removal of low-frequency noise: Butterworth highpass filters 1613.7.3 Removal of periodic artifacts: Notch and comb filters 1623.8 Order-statistic Filters 1693.9 The Wiener Filter 1713.10 Adaptive Filters for Removal of Interference 1803.10.1 The adaptive noise canceler 1813.10.2 The least-mean-squares adaptive filter 1843.10.3 The RLS adaptive filter 1853.11 Selecting an Appropriate Filter 1903.12 Application: Removal of Artifacts in ERP Signals 1933.13 Application: Removal of Artifacts in the ECG 1963.14 Application: Maternal–Fetal ECG 1973.15 Application: Muscle-contraction Interference 1993.16 Remarks 2023.17 Study Questions and Problems 2023.18 Laboratory Exercises and Projects 208References 2094 Detection of Events 2134.1 Problem Statement 2134.2 Illustration of the Problem with Case Studies 2144.2.1 The P, QRS, and T waves in the ECG 2144.2.2 The first and second heart sounds 2154.2.3 The dicrotic notch in the carotid pulse 2154.2.4 EEG rhythms, waves, and transients 2154.3 Detection of Events and Waves 2184.3.1 Derivative-based methods for QRS detection 2184.3.2 The Pan–Tompkins algorithm for QRS detection 2204.3.3 Detection of the P wave in the ECG 2244.3.4 Detection of the T wave in the ECG 2264.3.5 Detection of the dicrotic notch 2284.4 Correlation Analysis of EEG Rhythms 2284.4.1 Detection of EEG rhythms 2284.4.2 Template matching for EEG spike-and-wave detection 2314.4.3 Detection of EEG rhythms related to seizure 2344.5 Cross-spectral Techniques 2354.5.1 Coherence analysis of EEG channels 2354.6 The Matched Filter 2374.6.1 Derivation of the transfer function of the matched filter 2374.6.2 Detection of EEG spike-and-wave complexes 2414.7 Homomorphic Filtering 2424.7.1 Generalized linear filtering 2444.7.2 Homomorphic deconvolution 2444.7.3 Extraction of the vocal-tract response 2454.8 Application: ECG Rhythm Analysis 2534.9 Application: Identification of Heart Sounds 2544.10 Application: Detection of the Aortic Component of S 2 2564.11 Remarks 2594.12 Study Questions and Problems 2594.13 Laboratory Exercises and Projects 261References 2625 Analysis of Waveshape and Waveform Complexity 2675.1 Problem Statement 2675.2 Illustration of the Problem with Case Studies 2685.2.1 The QRS complex in the case of bundle-branch block 2685.2.2 The effect of myocardial ischemia on QRS waveshape 2685.2.3 Ectopic beats 2685.2.4 Complexity of the EMG interference pattern 2685.2.5 PCG intensity patterns 2695.3 Analysis of ERPs 2695.4 Morphological Analysis of ECG Waves 2695.4.1 Correlation coefficient 2705.4.2 The minimum-phase correspondent and signal length 2705.4.3 ECG waveform analysis 2745.5 Envelope Extraction and Analysis 2775.5.1 Amplitude demodulation 2785.5.2 Synchronized averaging of PCG envelopes 2805.5.3 The envelogram 2815.6 Analysis of Activity 2835.6.1 The RMS value 2835.6.2 Zero-crossing rate 2855.6.3 Turns count 2855.6.4 Form factor 2865.7 Application: Normal and Ectopic ECG Beats 2875.8 Application: Analysis of Exercise ECG 2885.9 Application: Analysis of the EMG in Relation to Force 2905.10 Application: Analysis of Respiration 2925.11 Application: Correlates of Muscular Contraction 2945.12 Application: Statistical Analysis of VAG Signals 2955.12.1 Acquisition of knee-joint VAG signals 2975.12.2 Estimation of the PDFs of VAG signals 2975.12.3 Screening of VAG signals using statistical parameters 2995.13 Application: Fractal Analysis of the EMG in Relation to Force 3025.13.1 Fractals in nature 3025.13.2 Fractal dimension 3035.13.3 Fractal analysis of physiological signals 3045.13.4 Fractal analysis of EMG signals 3055.14 Remarks 3065.15 Study Questions and Problems 3075.16 Laboratory Exercises and Projects 309References 3106 Frequency-domain Characterization of Signals and Systems 3176.1 Problem Statement 3186.2 Illustration of the Problem with Case Studies 3186.2.1 The effect of myocardial elasticity on heart sound spectra 3186.2.2 Frequency analysis of murmurs to diagnose valvular defects 3196.3 Estimation of the PSD 3216.3.1 Considerations in the computation of the ACF 3216.3.2 The periodogram 3236.3.3 The need for averaging PSDs 3256.3.4 The use of windows: spectral resolution and leakage 3266.3.5 Estimation of the ACF from the PSD 3306.3.6 Synchronized averaging of PCG spectra 3316.4 Measures Derived from PSDs 3336.4.1 Moments of PSD functions 3346.4.2 Spectral power ratios 3376.5 Application: Evaluation of Prosthetic Heart Valves 3376.6 Application: Fractal Analysis of VAG Signals 3396.6.1 Fractals and the 1/f model 3396.6.2 F D via power spectral analysis 3416.6.3 Examples of synthesized fractal signals 3416.6.4 Fractal analysis of segments of VAG signals 3426.7 Application: Spectral Analysis of EEG Signals 3456.8 Remarks 3496.9 Study Questions and Problems 3506.10 Laboratory Exercises and Projects 351References 3537 Modeling of Biomedical Signal-generating Processes and Systems 3577.1 Problem Statement 3577.2 Illustration of the Problem 3587.2.1 Motor-unit firing patterns 3587.2.2 Cardiac rhythm 3587.2.3 Formants and pitch in speech 3597.2.4 Patellofemoral crepitus 3607.3 Point Processes 3607.4 Parametric System Modeling 3657.5 Autoregressive or All-pole Modeling 3697.5.1 Spectral matching and parameterization 3747.5.2 Optimal model order 3777.5.3 AR and cepstral coefficients 3847.6 Pole–Zero Modeling 3847.6.1 Sequential estimation of poles and zeros 3877.6.2 Iterative system identification 3897.6.3 Homomorphic prediction and modeling 3937.7 Electromechanical Models of Signal Generation 3957.7.1 Modeling of respiratory sounds 3967.7.2 Modeling sound generation in coronary arteries 4007.7.3 Modeling sound generation in knee joints 4027.8 Electrophysiological Models of the Heart 4047.8.1 Electrophysiological modeling at the cellular level 4057.8.2 Electrophysiological modeling at the tissue and organ levels 4107.8.3 Extensions to the models of the heart 4127.8.4 Challenges and future considerations in modeling the heart 4147.9 Application: Heart-rate Variability 4167.10 Application: Spectral Modeling and Analysis of PCG Signals 4187.11 Application: Coronary Artery Disease 4217.12 Remarks 4237.13 Study Questions and Problems 4247.14 Laboratory Exercises and Projects 425References 4268 Adaptive Analysis of Nonstationary Signals 4318.1 Problem Statement 4328.2 Illustration of the Problem with Case Studies 4328.2.1 Heart sounds and murmurs 4328.2.2 EEG rhythms and waves 4338.2.3 Articular cartilage damage and knee-joint vibration 4338.3 Time-variant Systems 4358.3.1 Characterization of nonstationary signals and dynamic systems 4368.4 Fixed Segmentation 4388.4.1 The short-time Fourier transform 4388.4.2 Considerations in short-time analysis 4418.5 Adaptive Segmentation 4458.5.1 Spectral error measure 4458.5.2 ACF distance 4508.5.3 The generalized likelihood ratio 4508.5.4 Comparative analysis of the ACF, SEM, and GLR methods 4528.6 Use of Adaptive Filters for Segmentation 4528.6.1 Monitoring the RLS filter 4538.6.2 The RLS lattice filter 4568.7 The Kalman Filter 4638.8 Wavelet Analysis 4748.8.1 Approximation of a signal using wavelets 4748.9 Bilinear TFDs 4798.10 Application: Adaptive Segmentation of EEG Signals 4858.11 Application: Adaptive Segmentation of PCG Signals 4898.12 Application: Time-varying Analysis of HRV 4908.13 Application: Analysis of Crying Sounds of Infants 4938.14 Application: Wavelet Denoising of PPG Signals 4938.15 Application: Wavelet Analysis for CPR Studies 4948.16 Application: Detection of Ventricular Fibrillation in ECG Signals 4998.17 Application: Detection of Epileptic Seizures in EEG Signals 5038.18 Application: Neural Decoding for Control of Prostheses 5058.19 Remarks 5068.20 Study Questions and Problems 5078.21 Laboratory Exercises and Projects 507References 5089 Signal Analysis via Adaptive Decomposition 5159.1 Problem Statement 5179.2 Illustration of the Problem with Case Studies 5179.2.1 Separation of the fetal ECG from a single-channel abdominal Ecg 5179.2.2 Patient-specific EEG channel selection for BCI applications 5189.2.3 Detection of microvolt T-wave alternans in long-term ECG recordings 5189.3 Matching Pursuit 5189.4 Empirical Mode Decomposition 5209.4.1 Variants of empirical mode decomposition 5219.5 Dictionary Learning 5239.6 Decomposition-based Adaptive TFD 5259.7 Separation of Mixtures of Signals 5319.7.1 Principal component analysis 5339.7.2 Independent component analysis 5399.7.3 Nonnegative matrix factorization 5429.7.4 Comparison of PCA, ICA, and NMF 5469.8 Application: Detection of Epileptic Seizures Using Dictionary Learning Methods 5539.9 Application: Adaptive Time–Frequency Analysis of VAG Signals 5609.10 Application: Detection of T-wave Alternans in ECG Signals 5689.11 Application: Extraction of the Fetal ECG from Single-channel Maternal ECG 5729.12 Application: EEG Analysis for Brain–Computer Interfaces 5779.12.1 NMF-based channel selection 5799.12.2 Feature extraction 5799.13 Remarks 5869.14 Study Questions and Problems 5869.15 Laboratory Exercises and Projects 586References 58710 Computer-aided Diagnosis and Healthcare 59510.1 Problem Statement 59610.2 Illustration of the Problem with Case Studies 59610.2.1 Diagnosis of bundle-branch block 59610.2.2 Normal or ectopic ECG beat? 59710.2.3 Is there an alpha rhythm? 59810.2.4 Is a murmur present? 59810.2.5 Detection of sleep apnea using multimodal biomedical signals 59810.3 Pattern Classification 59910.4 Supervised Pattern Classification 60010.4.1 Discriminant and decision functions 60010.4.2 Fisher linear discriminant analysis 60110.4.3 Distance functions 60510.4.4 The nearest-neighbor rule 60510.4.5 The support vector machine 60610.5 Unsupervised Pattern Classification 60710.5.1 Cluster-seeking methods 60710.6 Probabilistic Models and Statistical Decision 61110.6.1 Likelihood functions and statistical decision 61110.6.2 Bayes classifier for normal patterns 61310.7 Logistic Regression Analysis 61410.8 Neural Networks 61510.8.1 ANNs with radial basis functions 61710.8.2 Deep learning 62010.9 Measures of Diagnostic Accuracy and Cost 62010.9.1 Receiver operating characteristics 62310.9.2 McNemar’s test of symmetry 62510.10 Reliability of Features, Classifiers, and Decisions 62710.10.1 Separability of features 62810.10.2 Feature selection 63010.10.3 The training and test steps 63110.11 Application: Normal versus Ectopic ECG Beats 63310.11.1 Classification with a linear discriminant function 63310.11.2 Application of the Bayes classifier 63710.11.3 Classification using the K-means method 63710.12 Application: Detection of Knee-joint Cartilage Pathology 63710.13 Application: Detection of Sleep Apnea 64410.14 Application: Monitoring Parkinson’s Disease Using Multimodal Signal Analysis 64710.15 Strengths and Limitations of CAD 65010.16 Remarks 65610.17 Study Questions and Problems 65710.18 Laboratory Exercises and Projects 658References 659Index 665