Signal Processing Strategies

Neural engineering is an emerging and fast-moving interdisciplinary research area that combines engineering with (a) electronic and photonic technologies, (b) computer science, (c) physics, (d) chemistry, (e) mathematics, and (f) cellular, molecular, cognitive, and behavioral neuroscience. This helps us understand the organizational principles and underlying mechanisms of the biology of neural systems and to further to study the behavioral dynamics and complexity of neural systems in nature. The field of neural engineering deals with many aspects of basic and clinical problems associated with neural dysfunction, including (i) the representation of sensory and motor information, (ii) electrical stimulation of the neuromuscular system to control muscle activation and movement, (iii) the analysis and visualization of complex neural systems at multiscale from the single cell to system levels to understand the underlying mechanisms, (iv) development of novel electronic and photonic devices and techniques for experimental probing, the neural simulation studies, (v) the design and development of human–machine interface systems and artificial vision sensors, and (vi) neural prosthesis to restore and enhance the impaired sensory and motor systems and functions. To highlight this emerging discipline, Dr. Ayman El-Baz and Dr. Jasjit Suri have developed Advances in Neural Engineering, covering the broad spectrum of neural engineering subfields and applications. This Series includes 7 volumes in the following order: Volume 1: Signal Processing Strategies, Volume 2: Brain-Computer Interfaces, Volume 3: Diagnostic Imaging Systems, Volume 4: Brain Pathologies and Disorders, Volume 5: Computing and Data Technologies, Volume 6: Advanced Brain Imaging Techniques and Volume 7: Neural Science Ethics. Volume 1 provides a comprehensive review of dominant feature extraction methods and classification algorithms in the brain-computer interfaces for motor imagery tasks. The authors discuss existing challenges in the domain of motor imagery brain-computer interface and suggest possible research directions.

Title of Book
Cover image
Title page
Table of Contents
Front matter
Titles in this Series
Series Editors
Copyright
Contributors
1. Framework for segmentation, optimization, and recognition of multivariate brain tumors
1 Introduction
2 Literature review
3 Background
3.1 Segmentation
3.2 Convolutional neural network
3.3 Parameters optimization
3.4 Transfer learning
3.5 Data augmentation
3.6 Metaheuristic optimization
3.6.1 Sparrow search algorithm
3.7 Performance metrics
4 Methodology
4.1 Dataset acquisition and preprocessing
4.1.1 Datasets acquisition
4.1.2 Datasets preprocessing
4.2 Segmentation phase
4.2.1 The U-Net model
4.2.2 The U-Net++ model
4.2.3 The attention U-Net model
4.2.4 The V-Net model
4.3 Learning and optimization
4.3.1 Initialization
4.3.2 Objective function calculation
4.3.3 Population sorting
4.3.4 Selection
4.3.5 Population updating
4.4 The overall pseudocode
5 Experiments and discussions
5.1 Experiments configurations
5.2 Segmentation experiments
5.3 Random hyperparameters experiments
5.3.1 Two-classes dataset experiments
5.3.2 Four-classes dataset experiments
5.4 Learning and optimization experiments
5.4.1 Two-classes dataset experiment
5.4.2 Four-classes dataset experiment
6 Conclusion, limitation, and future work
2. The neural circuitry of PTSD—An RDOC approach
1 Introduction
2 Pathophysiology of PTSD
2.1 Negative valence systems
2.2 Positive valence systems
2.3 Cognitive systems
2.4 Systems for social processes
2.5 Arousal symptoms
3 Neurotechnological strategies towards PTSD treatment
3.1 Neurofeedback (nFb)
3.2 Transcranial magnetic stimulation (TMS)
3.3 Transcranial direct current stimulation (tDCS)
3.4 Deep brain stimulation (DBS)/Responsive neurostimulation (RNS)
4 Conclusion
3. CNN-based artifact recognition from independent components of EEG signals
1 Introduction
2 ICA-based artifact removal pipeline
2.1 ICA calculation
2.2 Topoplot generation
2.3 Artifact recognition
2.3.1 Software tools for human experts
2.3.2 Automatic ICA-based strategies using topoplots
3 Report on a case of study method
3.1 The architecture
3.2 The experimental data set
3.3 Data filtering and augmentation
3.4 Data set distribution and training
3.5 Results
4 Conclusions
4. Deep multimodal representation learning for noninvasive neural speech decoding
1 Introduction
2 Multimodal decoding of overt and imagined speech
2.1 Study population
2.2 Experimental protocol
2.3 Data acquisition and signal processing
2.4 Multimodal neural network
2.4.1 Convolution
2.4.2 Non-linear activation
2.4.3 Batch normalization and dropout
2.4.4 Multimodal neural network implementation
2.4.5 Unimodal neural network
2.4.6 Training procedure
2.4.7 Hyperparameter optimization
3 Results
3.1 Multimodal learning outperforms unimodal
3.2 Decoding performance of the multimodal approach
3.3 Stimulus effects greater than word-type
4 Discussion
5 Conclusion
5. Neural signals processing using deep learning for diagnosis of cognitive disorders
1 Introduction
2 Deep learning-aided medical diagnosis systems
2.1 Types of neural signals
2.2 Paradigms for neural signals processing and analytics
2.3 Feature extraction techniques
2.4 Deep learning essentials
2.4.1 Artificial neural networks
2.4.2 Deep learning architectures
2.4.3 Training deep neural networks
2.4.4 Deep learning applications
2.5 DL models for neural signal analytics
2.5.1 Deep convolutional neural networks
2.5.2 Recurrent neural networks for EEG and ECoG analysis
2.5.3 Generative adversarial networks for brain image synthesis
2.5.4 Deep reinforcement learning for neurofeedback and brain-controlled interfaces
2.6 Clinical application
3 Case study: Diagnosis of ADHD
3.1 Introduction to ADHD
3.2 Study population and dataset
3.2.1 Subjects
3.3 Feature extraction
3.3.1 Data epochs
3.4 Model architecture
3.4.1 Convolutional neural network
3.4.2 Modified ResNet neural network
3.5 Experimental results
4 Discussion
5 Conclusion
6. Brain tumor recognition using semisupervised generative adversarial network
1 Introduction
2 Proposed methodology
2.1 Preprocessing
2.2 Classification using semi supervised generative adversarial network
3 Experimentations and results
3.1 Dataset
3.2 Preprocessing
3.3 Performance evaluation of SSGAN
4 Analysis
5 Conclusions and future scopes
7. Multivariate adaptive signal decomposition techniques and their applications to EEG signal processing: An introduction
1 Introduction
2 Multivariate time series
3 Multivariate adaptive decomposition
3.1 Multivariate empirical mode decomposition
3.2 Multivariate empirical wavelet transform
3.3 Multivariate Fourier-Bessel series expansion-based empirical wavelet transform
3.4 Multivariate variational mode decomposition
3.5 Multivariate iterative filtering
3.6 Other multivariate adaptive decomposition techniques
4 Application of multivariate adaptive decomposition to EEG signal processing
4.1 Brain-computer interface (BCI)
4.2 Neurological disease diagnosis
5 Conclusion
8. Split learning for human activity recognition
1 Introduction
2 Related work
3 Data and preprocessing
4 Methodology
4.1 Feature extraction
4.2 Model architecture
4.3 Split learning setup
5 Experimental setup
5.1 Evaluation setup
5.2 Metrics and scenarios
6 Results and discussion
7 Conclusion
9. Machine learning approaches for epilepsy analysis in current clinical trials
1 Introduction
2 Background studies
2.1 Types of epilepsy
2.2 Dataset description for epilepsy analysis
2.2.1 EEG database
2.2.2 Neuroimaging
2.3 Preprocessing for epilepsy analysis
2.4 Feature extraction and selection for epilepsy analysis
2.5 Predictive models for epilepsy analysis
2.6 Performance evaluation for epilepsy analysis
3 Machine learning techniques for epilepsy analysis
3.1 Predictive and classification modeling
3.1.1 Logistic regression
3.1.2 K-nearest neighbors (k-NN)
3.1.3 Decision trees
3.1.4 Naive Bayes
3.1.5 Support vector machines (SVM)
3.1.6 Random forests
3.1.7 Neural networks
4 Challenges and considerations in clinical trials
5 Case studies: Machine learning in current clinical trials
5.1 Case study 1: Real-time seizure prediction using EEG data
5.2 Case study 2: Predictive modeling of treatment response in epilepsy patients
5.3 Case study 3: Predicting drug responsiveness in epilepsy patients
5.4 Case study 4: Deep learning-based brain lesion segmentation in MRI
5.5 Case study 5: Closed-loop deep brain stimulation for seizure control
6 Future directions and implications
6.1 Current trends and emerging technologies
6.2 Advancements in data collection and integration
6.3 Potential impact on clinical practice
6.4 Ethical considerations and patient-centric approaches
7 Conclusion
10. Brainwave and head motion control of a smart home for disabled people
1 Introduction
2 Proposed methodology
2.1 EEG signal extraction
2.2 Android application development
2.3 Control circuit development
2.4 EEG signal processing
2.5 Subsystems integration
3 Results and analysis
3.1 Attention level data analysis
3.2 Blinking data analysis
3.3 Motion data analysis
3.4 Appliances control
4 Conclusion
11. Independent component analysis methods for motor imagery-based brain-computer interfaces
1 Introduction
2 Methods
2.1 Dataset
2.2 Source extraction
2.3 Feature extraction and selection
2.4 Classification
2.5 EEGNet
3 Results
3.1 ICA and classifier comparison
3.2 BCI Competition IV 2a
3.2.1 The best combination per subject
3.2.2 The best ICA - per subject
3.2.3 The best classifier - per subject
3.2.4 The best classifier per ICA - average over the dataset
3.2.5 The best ICA per classifier - average over the dataset
3.2.6 The best ICA and the best classifier - average over the dataset
3.3 OpenBMI
3.3.1 The best classifier per ICA - average over the dataset
3.3.2 The best ICA per classifier - average over the dataset
3.3.3 The best ICA and best classifier - average over the dataset
4 EEGNet
4.1 The best ICA - per subject
4.2 Training convergence speed per ICA
5 Conclusion
5.1 Future work
12. Advancing neural engineering: Hierarchical control strategies with human-centered focus for hand prosthetics
1 Introduction
1.1 A neurological background on the motor control in the human body
2 Hierarchical control strategies in hand prosthesis
2.1 Types of hand prosthesis
2.1.1 Passive hand prosthesis
2.1.2 Active hand prosthesis
2.2 Need for hierarchical control approach in hand prosthesis
2.3 Hierarchy explored in control approaches for hand prosthesis
2.3.1 Myoelectric control
2.3.2 Brain-computer interfaces and neural control
2.3.3 Collaborative and shared autonomy control
2.4 Signal processing techniques in hand prosthesis
3 Sensory feedback in hierarchical control
3.1 Types of sensory feedback modalities
3.1.1 Electro-tactile feedback
3.1.2 Vibrotactile feedback
3.1.3 Visual feedback
3.2 Sensor technologies for sensory feedback
3.2.1 EMG sensors
3.2.2 EEG sensors
3.2.3 Force and pressure sensors
3.2.4 Tactile sensors
3.2.5 Position and inertial sensors
3.2.6 Computer vision and camera sensors
3.2.7 Vibration sensors
4 Challenges and future directions
5 Conclusion
13. Advances in non-invasive EEG-based brain-computer interfaces: Signal acquisition, processing, emerging approaches, and applications
1 Introduction
2 Signal acquisition techniques
2.1 Traditional wet electrodes: Principles and limitations
2.2 Advancements in dry electrodes
2.3 Introduction to alternative electrode technologies
2.4 Exploring electrode placement methods for improved signal acquisition
3 Electroencephalography (EEG) and signal processing techniques
3.1 Signal processing for enhancing EEG data
3.1.1 Noise reduction
3.1.2 Feature extraction
3.1.3 Analyzing brain connectivity and network dynamics
3.1.4 Coherence analysis
3.1.5 Graph theory
3.1.6 Connectivity metrics
3.2 Signal pre-processing methods for noise reduction and artifact removal
3.3 Feature extraction and feature selection techniques
3.4 Classification algorithms for decoding brain signals
4 Existing approaches, packages, datasets, and applications of EEG signal processing
4.1 Common Spatial Pattern
4.2 Filter-based approaches
4.2.1 Filter bank CSP (FBCSP)
4.2.2 Improved discriminative FBCSP (iDFBCSP)
4.2.3 Optimizing temporal filter parameters
4.3 Riemannian manifolds and tangent space mapping
4.4 Deep learning approaches for EEG signal processing
4.4.1 Optimized CSP and LSTM based predictor (OPTICAL)
4.4.2 Convolutional neural network approaches
4.5 Datasets, available packages, and performance comparison
4.6 Applications
5 Conclusion and future perspectives
Index

Life Sciences

Physical Sciences & Engineering

Social Sciences & Humanities

Health

Signal Processing Strategies

Purchase options

World Book Day Celebration!

Institutional subscription on ScienceDirect

Ayman S. El-Baz

Jasjit S. Suri

Related books