Augmenting Neurological Disorder Prediction and Rehabilitation Using Artificial Intelligence

Description

Augmenting Neurological Disorder Prediction and Rehabilitation Using Artificial Intelligence focuses on how the neurosciences can benefit from advances in AI, especially in areas such as medical image analysis for the improved diagnosis of Alzheimer’s disease, early detection of acute neurologic events, prediction of stroke, medical image segmentation for quantitative evaluation of neuroanatomy and vasculature, diagnosis of Alzheimer’s Disease, autism spectrum disorder, and other key neurological disorders. Chapters also focus on how AI can help in predicting stroke recovery, and the use of Machine Learning and AI in personalizing stroke rehabilitation therapy. Other sections delve into Epilepsy and the use of Machine Learning techniques to detect epileptogenic lesions on MRIs and how to understand neural networks. Front Cover Augmenting Neurological Disorder Prediction and Rehabilitation Using Artificial Intelligence Copyright Page Contents List of contributors 1 Intracranial hemorrhage detection and classification using deep learning 1.1 Introduction 1.2 Types of intracranial hemorrhage 1.2.1 Intraparenchymal hemorrhage 1.2.2 Intraventricular hemorrhage 1.2.3 Subarachnoid hemorrhage 1.2.4 Subdural hematoma 1.2.5 Epidural hematoma 1.3 Related work 1.4 Characteristic challenges of intracranial hemorrhage detection from computerized tomography images 1.4.1 The computerized tomography imaging process 1.4.2 Availability of data 1.4.3 Digital imaging and communications in medicine anonymization 1.4.4 Hounsfield values and windowing 1.5 Our approach 1.5.1 Dataset 1.5.2 Experiments 1.5.2.1 Data preprocessing 1.5.2.2 Windowing 1.5.2.3 Multilabel classification 1.5.2.4 Network architectures used 1.5.3 Results 1.6 Conclusion References 2 Deep learning for noninvasive management of brain tumors 2.1 Introduction 2.2 Related works 2.3 Foundation of deep convolutional neural networks 2.4 Brain tumors and multimodal magnetic resonance imaging 2.5 Multiplanar convolutional neural networks for volumetric brain tumor segmentation 2.6 Deep radiomic for brain tumor classification 2.6.1 Three level architectures 2.7 Experimental results 2.7.1 Multiplanar convolutional neural networks 2.7.2 Deep radiomic 2.8 Conclusion References 3 Artificial intelligence in Parkinson’s disease—symptoms identification and monitoring 3.1 Introduction 3.2 Materials and methods 3.2.1 Motor fluctuations identification and ON versus OFF discrimination 3.2.2 Dyskinesia 3.2.3 Bradykinesia 3.2.4 Tremor 3.2.5 Freezing of gait 3.3 Discussion References 4 Alzheimer’s disease detection using artificial intelligence 4.1 Introduction 4.2 Background/literature review 4.3 Classification methods for alzheimer’s disease detection 4.3.1 Deep learning 4.3.2 Convolutional neural networks 4.3.3 Transfer learning 4.4 Alzheimer’s disease detection using artificial intelligence 4.4.1 Experimental data 4.4.2 Performance evaluation measures 4.4.3 Experimental results 4.4.3.1 Transfer learning results 4.4.3.2 Results of convoluted neural network with different number of Layers 4.5 Discussion 4.6 Conclusion References 5 Intelligent computer systems for multiple sclerosis diagnosis 5.1 Introduction 5.2 A review of reasoning methods in intelligent systems for MS diagnosis 5.2.1 Overview 5.2.2 Methods 5.2.3 Results 5.2.4 Discussion 5.3 Rule-based CDSS for MS (relapsing-remitting) diagnosis 5.3.1 Overview of rule-based intelligent systems 5.3.2 Material and methods 5.3.3 Results 5.3.4 Discussion 5.4 Fuzzy rule-based CDSS for MS (relapsing-remitting) diagnosis 5.4.1 Overview of fuzzy rule-based intelligent systems 5.4.2 Material and methods 5.4.3 Result 5.4.3.1 Knowledge base of system 5.4.3.2 Input and output fuzzy variables and membership functions 5.4.3.3 Fuzzy rules 5.4.3.4 Inference engine 5.4.3.5 User interface system 5.4.3.6 System evaluation 5.4.4 Discussion 5.5 Conclusion References 6 Current and future applications of artificial intelligence in multiple sclerosis 6.1 Introduction 6.2 Artificial intelligence techniques in multiple sclerosis: machine learning and deep learning 6.3 Artificial intelligence on magnetic resonance imaging 6.3.1 Magnetic resonance imaging protocol improvement 6.3.2 Image analysis 6.3.3 Data harmonization 6.4 Artificial intelligence on other measures 6.4.1 Neurophysiological measures 6.4.2 Laboratory tests 6.5 Clinical applications 6.5.1 Diagnosis and prognosis in clinically isolated syndrome 6.5.2 Differential diagnosis 6.5.3 Prognosis and disease monitoring 6.5.4 Understanding disease pathophysiology 6.6 Future development 6.6.1 Solving the “black-box problem” 6.6.2 Towards personalized medicine 6.7 Conclusion References 7 Artificial intelligence–assisted headache classification: a review 7.1 Introduction 7.1.1 Artificial intelligence, machine learning, and deep learning 7.1.2 Different categories of machine learning algorithms 7.1.3 Different categories of deep learning algorithms 7.1.4 Headache 7.2 AI-based techniques for diagnosis, classification, and management of headache disorders 7.2.1 Studies using ML 7.2.2 Studies using DL 7.2.3 Expert/fuzzy systems 7.2.4 Biologically inspired algorithm 7.2.5 Decision support system 7.2.6 Natural language processing 7.3 Major drawbacks and challenges 7.4 Conclusion References 8 Deep learning for reliable detection of epileptogenic lesions 8.1 Background 8.2 Deep learning: a brief introduction 8.3 Deep learning is well-suited to detect common brain lesions 8.4 The challenge: how to detect rare brain lesions with deep learning? 8.5 Explainable and interpretable deep learning: a view into the black box 8.6 Generalizability of deep learning models 8.7 Conclusion Acknowledgments References 9 Artificial intelligence in neurosciences—are we really there? 9.1 Introduction 9.2 Exponential increase in healthcare artificial intelligence publications 9.3 Components of artificial intelligence 9.4 Overview of artificial intelligence in clinical neurosciences 9.5 Challenges in artificial intelligence implementation 9.6 Financial implications 9.7 Trust 9.8 Ethical challenges 9.9 Regulatory issues 9.10 Legal issues 9.11 Neuroscience and artificial intelligence 9.12 Conclusion Acknowledgment References 10 Artificial intelligence in the management of neurological disorders: its prevalence and prominence 10.1 Introduction 10.2 Overview of artificial, machine learning, and deep learning in healthcare 10.3 Type of data used by artificial intelligence in neurology 10.4 Artificial intelligence in neurology 10.4.1 Diagnostic classification and prediction of neurological disorders 10.4.1.1 Neuroradiology 10.4.1.2 Neurooncology 10.4.1.3 Neurodegenerative diseases 10.4.1.4 Cerebrovascular/neurovascular conditions 10.4.1.5 Traumatic brain injury 10.4.1.6 Spinal cord injury 10.4.1.7 Neurological/neonatal intensive care unit 10.4.2 Outcome prediction and prognosis for neurological disorders 10.4.3 Clinical interventions for neurological disorders 10.4.3.1 Clinical prospects of artificial intelligence-based neurology 10.4.3.2 Artificial intelligence in neurosurgery and surgical planning 10.5 Discussion: findings and open issues 10.5.1 Lack of large curated and annotated medical image data for training 10.5.2 Neuroprivacy 10.5.3 Variety of imaging protocols 10.5.4 Multimodal data 10.6 Conclusion References 11 Graphical assessment of the internal structure of Parkinsons dataset—a case study 11.1 Introduction to multivariate data analysis 11.2 The data on Parkinsons disease 11.2.1 An overall description of the data 11.2.2 Boxplots of individual vocal measurements 11.3 The biplot methodology 11.3.1 Matrix approximation by lower rank matrices 11.3.2 The concept of a biplot 11.3.3 Formal properties of a biplot for (nearly) rank 2 matrices 11.4 Analyzing Parkinson’s disease data using biplots 11.5 Selection of the classifier 11.5.1 First approach: using all 22 variables 11.5.2 Analysis of the subset V4–V6,V13, V23 11.5.3 Approaches by other authors 11.6 Discussion and closing remarks References 12 Applications of artificial intelligence to neurological disorders: current technologies and open problems 12.1 Introduction 12.2 Neurological disorders 12.2.1 A classification of neurological disorders 12.2.2 Challenges posed by neurological disorders 12.2.3 Informative data types for existing challenges 12.3 Artificial intelligence/machine learning algorithms 12.3.1 State-of-the-art approaches for analyzing different types of data 12.3.2 Applications of artificial intelligence/machine learning algorithms to neurological disorders 12.3.3 Open problems 12.4 Conclusion 12.5 Abbreviations References 13 Developing a chatbot/intelligent system for neurological diagnosis and management 13.1 Introduction 13.2 History of chatbot/intelligent system use in neurology 13.3 Different modalities of chatbot/intelligent systems in neurological diagnosis 13.4 Where is the intelligence in a chatbot/intelligent systems? 13.4.1 Designing intelligence 13.5 Designing a chatbot for a neurological condition 13.5.1 Framework and functionality 13.5.2 Snippets for pattern recognition 13.5.3 Dataset and threshold value assessment 13.5.4 Engine design 13.5.5 Experience design 13.6 Limitations of chatbots and intelligent systems 13.7 Ethical and legal issues 13.7.1 Biases 13.7.2 Data protection and confidentiality 13.7.3 Informed consent 13.7.4 Effectiveness, safety, and liability 13.7.5 Addressing ethical and legal issues 13.8 Future directions Acknowledgment References 14 Artificial intelligence in the diagnosis and management of acute ischemic stroke 14.1 Introduction 14.2 Different artificial intelligence types and their implication on stroke triage and management 14.3 Use of artificial intelligence in acute ischemic stroke 14.3.1 Detection 14.3.1.1 Tissue detection 14.3.1.2 Large vessel occlusion detection 14.3.2 Prediction of tissue outcome 14.3.3 Prediction of clinical outcome 14.4 Challenges in use of artificial intelligence in stroke care 14.4.1 Software validation and veracity 14.4.2 Informed consent 14.4.3 Accountability and expanding circle of carers 14.4.4 Safety, liability, and regulatory approvals 14.4.5 Integrated stroke services and corporate responsibility 14.4.6 Impact of artificial intelligence in stroke service 14.5 Using artificial intelligence in acute ischemic stroke—author experience 14.6 Conclusion and the way forward References 15 Towards intelligent extended reality in stroke rehabilitation 15.1 Introduction 15.2 Current methods of stroke rehabilitation 15.3 Current extended reality technologies in stroke rehabilitation 15.3.1 Motor rehabilitation 15.3.2 Speech therapy 15.3.3 Cognitive rehabilitation 15.3.4 Machine learning for improving extended reality 15.4 Strategies to improve immersion and presence 15.4.1 Data augmentation 15.4.2 Reinforcement learning 15.4.3 Algorithms and software implementations 15.4.4 Conversational Agents 15.4.5 Limitations 15.4.6 Ethical issues 15.5 Conclusions and future direction References Index Back Cover