Machine Learning and Deep Learning Techniques for Medical Science

The application of machine learning is growing exponentially into every branch of business and science, including medical science. This book presents the integration of machine learning (ML) and deep learning (DL) algorithms that can be applied in the healthcare sector to reduce the time required by doctors, radiologists, and other medical professionals for analyzing, predicting, and diagnosing the conditions with accurate results. The book offers important key aspects in the development and implementation of ML and DL approaches toward developing prediction tools and models and improving medical diagnosis. The contributors explore the recent trends, innovations, challenges, and solutions, as well as case studies of the applications of ML and DL in intelligent system-based disease diagnosis. The chapters also highlight the basics and the need for applying mathematical aspects with reference to the development of new medical models. Authors also explore ML and DL in relation to artificial intelligence (AI) prediction tools, the discovery of drugs, neuroscience, diagnosis in multiple imaging modalities, and pattern recognition approaches to functional magnetic resonance imaging images. This book is for students and researchers of computer science and engineering, electronics and communication engineering, and information technology; for biomedical engineering researchers, academicians, and educators; and for students and professionals in other areas of the healthcare sector. Presents key aspects in the development and the implementation of ML and DL approaches toward developing prediction tools, models, and improving medical diagnosis Discusses the recent trends, innovations, challenges, solutions, and applications of intelligent system-based disease diagnosis Examines DL theories, models, and tools to enhance health information systems Explores ML and DL in relation to AI prediction tools, discovery of drugs, neuroscience, and diagnosis in multiple imaging modalities Dr. K. Gayathri Devi is a Professor at the Department of Electronics and Communication Engineering, Dr. N.G.P Institute of Technology, Tamil Nadu, India. Dr. Kishore Balasubramanian is an Assistant Professor (Senior Scale) at the Department of EEE at Dr. Mahalingam College of Engineering & Technology, Tamil Nadu, India. Dr. Le Anh Ngoc is a Director of Swinburne Innovation Space and Professor in Swinburne University of Technology (Vietnam). Cover Half Title Series Page Title Page Copyright Page Contents Editor Biographies Contributors 1. A Comprehensive Study on MLP and CNN, and the Implementation of Multi-Class Image Classification using Deep CNN 1.1 Introduction 1.2 The Processes of the Neural Network 1.2.1 Basics of Neural Network 1.2.1.1 Architecture of Neural Network 1.2.1.2 Working Principles of Neural Network 1.2.1.3 Learning Methods of Neural Network 1.2.1.4 Drawbacks of Neural Network 1.2.2 Convolutional Neural Network (CNN) Algorithm 1.2.2.1 Merits of CNN over MLP 1.2.2.2 Contents of CNN 1.2.2.3 Working of CNN Algorithm 1.2.2.3.1 Convolution Layer Padding Striding 1.2.2.3.2 Pooling Layer Pooling Layer Types 1.2.2.3.3 Fully Connected Layer (FC) 1.2.2.3.4 Dropout 1.2.2.3.5 Activation Functions 1.2.2.4 Deep CNN 1.3 Experimental Procedure 1.3.1 Preparing the Dataset 1.3.2 Model Training and Testing 1.4 Results and Discussion 1.4.1 MNIST Dataset Image Classifications 1.4.2 CIFAR-10 Dataset Image Classifications 1.5 Conclusion References 2. An Efficient Technique for Image Compression and Quality Retrieval in Diagnosis of Brain Tumour Hyper Spectral Image 2.1 Introduction 2.2 Literature Survey 2.2.1 Proposed System 2.2.2 Data Latches with D-flip Flops 2.2.3 Discussion and Results of the Simulation 2.3 Conclusion References 3. Classification of Breast Thermograms using a Multi-layer Perceptron with Back Propagation Learning 3.1 Introduction 3.2 Related Works 3.3 Methods & Materials 3.3.1 Pre-processing of Thermograms & Region of Interest (ROI) Segmentation 3.3.2 Feature Extraction & Selection 3.3.3 Designing Steps of a Multi-Layer Perceptron with Back Propagation Learning 3.3.3.1 Phase I: Feedforward Computations 3.3.3.2 Phase II: Back Propagation of the Error 3.3.3.3 Phase III: Update Weights and Error in the Output and Hidden Units 3.4 Performance Evaluation Parameters 3.5 Classification Results & Discussion 3.5.1 ANN Model with 5 Neurons in Hidden Layer 3.5.2 ANN Model with 10 Neurons in Hidden Layer 3.5.3 ANN Model with 15 Neurons in Hidden Layer 3.6 Conclusion & Future Work References 4. Neural Networks for Medical Image Computing 4.1 Introduction 4.2 Structure of Neural Network 4.3 Learning Process in Neural Networks 4.3.1 Supervised Learning 4.3.1.1 An Overview of Supervised Learning 4.3.1.2 Supervised Learning in Medical Image Processing 4.3.2 Unsupervised Learning 4.3.2.1 Unsupervised Learning 4.3.2.2 Overview of Competitive Learning 4.3.2.3 Medical Analysis using Unsupervised Learning 4.3.3 Reinforcement Learning 4.3.3.1 Reinforcement Learning 4.3.3.2 Overview of Q-Learning 4.3.3.3 Adopting Reinforcement Learning in Health Sector 4.4 Types of Neural Networks 4.4.1 Perceptron 4.4.1.1 Perceptron 4.4.1.2 Perceptron in Medical Image Analysis 4.4.2 Radial Basis Function Network 4.4.2.1 Architecture of Radial Basis Function Network 4.4.2.2 Implementing Radial Basis Function in Medical Analysis 4.4.3 Convolutional Neural Network 4.4.3.1 Architecture of Convolutional Neural Network 4.4.3.2 Convolutional Neural Networks in Medical Diagnosis 4.4.4 Recurrent Neural Network 4.4.4.1 Introduction to Recurrent Neural Network 4.4.4.2 Types of Recurrent Neural Network 4.4.4.3 Medical Analysis using Recurrent Neural Network 4.4.5 Hopfield Neural Network 4.4.5.1 Overview of Hopfield Neural Network 4.4.5.2 Hopfield Neural Network in Medical Diagnosis 4.5 Conclusion References 5. Recent Trends in Bio-Medical Waste, Challenges and Opportunities 5.1 Introduction 5.2 Waste Disposal 5.3 Health Care Industries 5.4 Conclusion References 6. Teager-Kaiser Boost Clustered Segmentation of Retinal Fundus Images for Glaucoma Detection 6.1 Preamble 6.2 Methodology 6.2.1 Nonlinear Teager-Kaiser Filtering Technique 6.2.2 Teager-Kaiser Boost Clustered Segmentation 6.2.3 Clinical Feature Extraction 6.3 Results and Discussion 6.3.1 Quantitative Analysis 6.4 Conclusion References 7. IoT-Based Deep Neural Network Approach for Heart Rate and SpO2 Prediction 7.1 Introduction 7.1.1 Related Work 7.1.2 Motivations 7.2 Materials and Methods 7.2.1 Complete DNN-based System 7.2.2 Principal Component Analysis (PCA) 7.2.3 DNN Model 7.2.4 Cloud Computing 7.3 Results 7.3.1 DNN Model Accuracy Performance 7.3.2 System Validation 7.3.2.1 Bland-Altman Analysis 7.3.2.2 R2 (Coefficient of Determination) Regression Score Function 7.3.3 Performance Analysis Criteria 7.4 Discussion 7.5 Conclusion References 8. An Intelligent System for Diagnosis and Prediction of Breast Cancer Malignant Features using Machine Learning Algorithms 8.1 Introduction 8.2 Machine Learning Technologies 8.2.1 Naïve Bayes 8.2.2 K-Nearest Neighbor 8.2.3 Random Forest 8.2.4 Support Vector Machine 8.3 Related Work 8.4 Proposed Methodology 8.4.1 Experimental Results and Discussions 8.4.1.1 Efficiency 8.5 Conclusion References 9. Medical Image Classification with Artificial and Deep Convolutional Neural Networks: A Comparative Study 9.1 Introduction 9.2 Machine and Deep Learning Methods 9.2.1 Machine Learning Techniques 9.2.1.1 Supervised Learning 9.2.1.2 Unsupervised Learning 9.2.1.3 Semi-Supervised Learning 9.2.1.4 Reinforcement Learning 9.2.1.5 Deep Learning 9.2.2 Deep Learning Techniques 9.2.2.1 DL Definitions 9.2.2.2 DL Class 9.2.2.3 Deep Architectures 9.3 Comprehensive Study 9.3.1 Concept of Brain MRI Data 9.3.2 Image Classification for Medical Disease Diagnosis 9.3.3 Medical Image Classification for Machine and Deep Learning 9.4 Comparative Study 9.5 Artificial and Convolutional Deep Neural Networks based on Medical Image Classification for Alzheimer Disease 9.5.1 Brain MRI Datasets 9.5.2 MRI Data Pre-processing 9.5.3 Features Extraction and Selection from Brain MRI Datasets 9.5.4 Classification Methods 9.5.5 Proposed Machine-Deep Model 9.6 Discussion and Conclusion References 10. Convolutional Neural Network for Classification of Skin Cancer Images 10.1 Introduction 10.2 State-of-the-Art 10.3 Materials and Methods 10.3.1 Data Preprocessing and Augmentation 10.3.2 Data Augmentation 10.3.3 Classification Models 10.3.3.1 Convolutional Neural Network (CNN) 10.3.3.2 Transfer Learning and Pre-trained Models 10.3.3.3 Pre-trained Xception Model 10.3.3.4 Xception Model Fine-tuning 10.3.3.5 Evaluation Metrics 10.4 Experimental Results 10.4.1 Learning Performance 10.4.2 Classification Results 10.4.3 Comparative Study 10.5 Conclusion and Perspectives References 11. Application of Artificial Intelligence in Medical Imaging 11.1 Introduction 11.2 Machine Learning 11.2.1 Supervised Learning 11.2.2 Unsupervised Learning 11.2.3 Semi-supervised Learning 11.2.4 Active Learning 11.2.5 Reinforcement Learning 11.2.6 Evolutionary Learning 11.2.7 Deep Learning 11.3 Use of Machine Learning for Medical Imaging 11.4 Deep Learning in Medical Imaging 11.4.1 Image Categorisation 11.4.2 Object Classification 11.4.3 Organ or Region Detection 11.4.4 Data Mining 11.4.5 The Sign-up Process 11.5 Summary References 12. Machine Learning Algorithms Used in Medical Field with a Case Study 12.1 Introduction 12.2 Machine Learning Algorithms 12.2.1 Supervised Learning 12.2.2 Unsupervised Learning 12.2.3 Reinforcement Learning 12.2.4 Semi-Supervised Learning 12.2.5 Regression Algorithms 12.2.6 Instance-based Algorithms 12.2.7 Regularization Algorithms 12.2.8 Decision Tree Algorithms 12.2.9 Bayesian Algorithms 12.2.10 Clustering Algorithms 12.2.11 Association Rule Learning Algorithms 12.2.12 Artificial Neural Network Algorithms 12.2.13 Deep Learning Algorithms 12.2.14 Dimensionality Reduction Algorithms 12.2.15 Ensemble Algorithms 12.3 ML Algorithms in Medical Diagnosis 12.4 ML Classifiers in Breast Cancer Diagnosis 12.4.1 Logistic Regression 12.4.2 K-Nearest Neighbor (k-NN) Algorithm 12.4.3 Support Vector Machine 12.4.4 Random Forest Classifier 12.4.5 Naive Bayes Classifier 12.4.6 Decision Tree Classifiers 12.4.7 Dimensionality Reduction Algorithms 12.5 Materials and Methods 12.6 Conclusion References 13. Dual Customized U-Net-based Automated Diagnosis of Glaucoma 13.1 Introduction 13.2 Literature Review 13.3 Proposed Work 13.4 Performance Measures 13.5 Simulation Results 13.5.1 Optic Disc Segmentation 13.6 Conclusion References 14. MuSCF-Net: Multi-scale, Multi-Channel Feature Network Using Resnet-based Attention Mechanism for Breast Histopathological Image Classification 14.1 Introduction 14.2 Related Studies 14.3 Contribution 14.4 Material and Methods 14.4.1 BREAKHIS Database 14.4.2 Methodology 14.4.3 Preprocessing 14.4.3.1 Patch Creation 14.4.3.2 Augmentation 14.4.3.3 MuSCF-Net Mechanism 14.4.3.3.1 Convolution of filter bank for feature extraction 14.4.3.3.2 Convolution Block Attention Module (CBMA) Integrated with ResBlock in ResNet 14.4.3.3.2.1 CBMA Integrated with ResBlockin Resnet 14.4.3.3.2.2 Channel Attention (CA) Module 14.4.3.3.2.3 Spatial Attention (SA) Module 14.4.3.3.2.4 ResBlock in Resnet [24] 14.4.3.3.2.5 Global Average Pooling Layer 14.4.3.3.2.6 Dropout Layer 14.4.3.3.2.7 Dense Block 14.4.4 Training Details 14.4.4.1 Adam Optimizer [37] 14.4.4.2 Activation Function 14.4.4.3 ReLU 14.4.4.4 Softmax Activation Function 14.4.4.5 Loss 14.5 Results and Discussion 14.6 Conclusion References 15. Artificial Intelligence is Revolutionizing Cancer Research 15.1 Introduction 15.2 Development of Artificial Intelligence in Medical Research 15.3 AI in Different Cancer Treatment Modalities 15.3.1 Drug Development 15.3.2 Chemotherapy 15.3.3 Radiotherapy 15.3.4 Immunotherapy 15.3.5 Identifying Drug Targets 15.4 AI in Cancer Prediction at an Early Stage 15.5 Future Perspective in AI 15.6 Conclusion References 16. Deep Learning to Diagnose Diseases and Security in 5G Healthcare Informatics 16.1 Introduction 16.2 Key Types of Learning Methods Used to Solve 5G Problems 16.2.1 Supervised Learning 16.2.2 Unsupervised Learning 16.2.3 Reinforcement Learning 16.3 Main Deep Learning Techniques Used in 5G Scenarios 16.3.1 Fully Connected Models 16.3.2 Recurrent Neural Networks 16.3.3 CNN 16.3.4 DBN 16.3.5 Autoencoder 16.3.6 Combining Models 16.4 Most Common Scenarios Used for 5G Assessment and Deep Learning Integration 16.5 Applications of Machine Learning and Deep Learning for 5G Security 16.6 Blockchain Technology in Healthcare 16.7 Evolution of Machine Learning in Disease Detection 16.7.1 Supervised Learning 16.7.1.1 K-Nearest Neighbour (KNN) 16.7.1.2 Support Vector Machine (SVM) 16.7.1.3 Decision Trees (DTs) 16.7.1.4 Classification and Regression Trees (CARTs) 16.7.1.5 Logistic Regression (LR) 16.7.1.6 Random Forest Algorithm (RFA) 16.7.1.7 Naive Bayes (NB) 16.7.1.8 Artificial Neural Network (ANN) 16.7.2 Unsupervised Learning 16.7.3 Semi-supervised Learning 16.7.4 Evolutionary Learning 16.7.5 Active Learning 16.7.6 Reinforcement Learning 16.7.7 Ensemble Learning 16.7.8 Deep Learning 16.7.9 Transfer Learning 16.7.9.1 Feature Extraction 16.7.9.2 Fine-tuning 16.8 Applications of Deep Learning in Disease Diagnosis 16.8.1 ML/DL in Healthcare: The Large Picture 16.8.2 A Look at the Healthcare Applications of ML and DL 16.9 Deep Learning in Disease Diagnosis: To Save Lives and Cuts Treatment Costs 16.9.1 Breast Cancer 16.9.2 Early Detection of Melanoma: Skin Cancer 16.9.3 Lung Cancer 16.9.4 Testing for Diabetic Retinopathy 16.9.5 Assessment of Cardiac Hazard from ECG Data 16.9.6 Using CT Scans of the Head to Detect Strokes Early 16.10 Benefits of Deep Learning 16.11 Scope of Deep Learning Techniques for Disease Diagnosis 16.12 A Deep Learning-Based Approach to Detect Neurodegenerative Diseases: Multiclass Classification (Case Study-1) 16.12.1 Material and Methods 16.12.1.1 ADPP Dataset Description 16.12.1.2 VGG 19 Architecture 16.12.2 Results and Discussion of the Above Discussed Framework 16.13 Pneumonia Detection using Deep Learning (Case Study-2) 16.13.1 Methodology 16.13.1.1 Structural Units 16.13.1.2 Architecture of CovXNet 16.13.1.3 Stacking of Multiple Networks 16.13.1.4 Transfer Learning Method of CovXNet for New Corona Virus Data 16.13.1.5 Network Training and Optimisation 16.13.2 Results and Discussions 16.13.2.1 Datasets 16.13.2.2 Evaluation of Performance 16.14 Early Detection of Deep Learning-based Diabetic Retinopathy (Case Study-3) 16.14.1 Datasets Used 16.14.2 Metric Assessment 16.14.2.1 Quadratic Weighted Kappa (QWK) 16.14.2.2 Intuition of Cohen's Kappa 16.14.2.3 Quadratic Weight Intuition in Ordinary Classes — Quadratic Weighted Kappa (QWK) 16.14.3 Method 16.14.3.1 Image Pre-processing and Augmentations 16.14.3.2 Network Architecture 16.14.3.3 Training Process 16.14.4 Results 16.14.4.1 Model Evaluation on Test Data 16.14.4.2 Other Transfer Learning Models 16.15 Conclusion References 17. New Approaches in Machine-based Image Analysis for Medical Oncology 17.1 Introduction 17.2 Classical Methods 17.3 Machine Learning Methods in Oncology 17.3.1 Supervised Learning 17.3.1.1 Support Vector Machine 17.3.1.2 Logistic or Linear Regression (LR) 17.3.1.3 Decision Tree 17.3.1.4 Random Forest Algorithm (RF) 17.3.1.5 Naive Bayes 17.3.1.6 K-Nearest Neighbour 17.3.1.7 Artificial Neural Network 17.3.2 Unsupervised Learning 17.3.2.1 K-means Clustering 17.3.2.2 Principle Component Analysis 17.3.2.3 Independent Component Analysis 17.3.2.4 Autoencoders 17.3.2.5 Singular Value Decomposition 17.3.3 Reinforcement Learning (RL) 17.4 Application of ML in Oncology 17.4.1 Brain Oncology 17.4.2 Skin Oncology 17.4.3 Breast Cancer Prognosis Prediction 17.4.4 ML in Lung Oncology 17.4.5 Gastric Oncology 17.5 Discussion 17.5.1 ML Program Performance Analysis 17.5.2 Pros and Cons of ML Algorithm 17.5.3 ML in Cancer Staging 17.5.4 Predicting and Evaluating Treatment Response 17.6 Conclusion References 18. Performance Analysis of Deep Convolutional Neural Networks for Diagnosing COVID-19: Data to Deployment 18.1 Introduction 18.2 Literature Review 18.3 The Attributes of the Dataset and Visualizations to Interpret the Data 18.4 The Model Formulation to Classify the Data to Diagnose COVID-19 18.4.1 InceptionResNetV2 18.4.2 ResNet152V2 18.4.3 Xception 18.4.4 DenseNet201 18.5 Loss Function: Categorical Cross Entropy 18.6 Evaluation Metrics and Results 18.7 Model Deployment 18.7.1 Designing the Website 18.7.2 An Overview of Deployment 18.7.3 Working of Website 18.8 Conclusion References 19. Stacked Auto Encoder Deep Neural Network with Principal Components Analysis for Identification of Chronic Kidney Disease 19.1 Introduction 19.2 Methodology 19.2.1 Stacked Auto-encoder Deep Neural Network 19.2.2 Principal Component Analysis (PCA) 19.3 Result and Discussion 19.4 Conclusion References Index