Cover Image

Data fusion techniques and applications for smart healthcare

Medical data exists in several formats, from structured data and medical reports to 1D signals, 2D images, 3D volumes, or even higher dimensional data such as temporal 3D sequences.

Bibliographic Details
Corporate Author: ScienceDirect (Online service)
Other Authors: Singh, Amit Kumar, Berretti, Stefano
Format: eBook
Language:English
Published: [S.l.] : Academic Press, 2024.
Subjects:
Artificial intelligence > Medical applications.
Intelligence artificielle > Applications en médecine.
Electronic books.
Online Access:Connect to the full text of this electronic book
Table of Contents:
  • Front Cover
  • Data Fusion Techniques and Applications for Smart Healthcare
  • Copyright
  • Contents
  • List of contributors
  • Preface
  • 1 Introduction
  • 2 Summary of book chapters
  • 3 Conclusions
  • 1 Retinopathy screening from OCT imagery via deep learning
  • 1.1 Introduction
  • 1.1.1 Background and motivation
  • 1.1.2 Related works
  • 1.1.3 Key contributions
  • 1.2 Retinal OCT classification framework
  • 1.2.1 Dataset description
  • 1.2.2 Deep learning architecture
  • 1.2.3 Loss function
  • 1.2.4 Training performance
  • 1.3 Simulation results
  • 1.3.1 Performance metrics
  • 1.3.2 System specification
  • 1.3.3 Experimental results
  • 1.4 Discussion
  • 1.5 Conclusion and future scope
  • References
  • 2 Multisensor data fusion in Digital Twins for smart healthcare
  • 2.1 Introduction
  • 2.2 Literature review
  • 2.3 SVM optimization
  • 2.4 DSET analysis
  • 2.5 Construction of a multisensor medical data decision-level fusion DTs model based on the combination of improved SVM and ...
  • 2.6 Experimental verification
  • 2.7 Results
  • 2.8 Discussion
  • 2.9 Conclusion
  • References
  • 3 Deep learning for multisource medical information processing
  • 3.1 Introduction and motivation
  • 3.2 Background, definitions, and notations
  • 3.2.1 Data fusion
  • 3.2.2 Sources of healthcare information and the role of multimodal fusion
  • 3.3 Literature review and state-of-the-art
  • 3.4 Problem definition
  • 3.5 Proposed solution
  • 3.5.1 Classification of acoustic data
  • 3.5.2 Fundamentals of deep learning
  • 3.5.3 Deep learning-based image and acoustic data fusion
  • 3.5.4 Random forest for multimodal fusion
  • 3.5.5 Performance evaluation metrics
  • 3.6 Challenges and prospective opinion
  • 3.6.1 Imbalanced data
  • 3.6.2 Overfitting
  • 3.6.3 Interpretability of data
  • 3.6.4 Uncertainty scaling
  • 3.6.5 Model compression
  • 3.6.6 Other challenges.
  • 3.6.7 Prospective opinion
  • 3.7 Conclusion
  • References
  • 4 Robust watermarking algorithm based on multimodal medical image fusion
  • 4.1 Introduction
  • 4.2 Literature survey
  • 4.3 Background information
  • 4.3.1 NSST
  • 4.3.2 IWT
  • 4.3.3 QR decomposition
  • 4.3.4 SVD
  • 4.4 Proposed methodology
  • 4.4.1 Multimodal image fusion
  • 4.4.2 Embedding and recovery procedure of fused mark image
  • 4.4.3 Encryption procedure
  • 4.4.4 Decryption process
  • 4.5 Results
  • 4.5.1 Objective analysis (PSNR-SSIM-NC)
  • 4.5.2 Robustness analysis
  • 4.5.3 Security analysis
  • 4.5.3.1 Statistical analysis
  • Histogram and chi-square analysis
  • Correlation analysis
  • Entropy analysis
  • 4.5.3.2 Differential analysis
  • 4.5.3.3 Key space and sensitivity analysis
  • 4.5.3.4 Time cost analysis
  • 4.6 Conclusion
  • Acknowledgments
  • References
  • 5 Fusion-based robust and secure watermarking method for e-healthcare applications
  • 5.1 Introduction
  • 5.2 Literature review
  • 5.3 Proposed method
  • 5.3.1 Watermark generation using image fusion
  • 5.3.2 Imperceptible and robust embedding of the fused image
  • 5.3.3 Encryption of the final watermarked image
  • 5.4 Results and analysis
  • 5.5 Conclusion
  • Acknowledgment
  • References
  • 6 Recent advancements in deep learning-based remote photoplethysmography methods
  • 6.1 Introduction
  • 6.2 Photoplethysmography
  • 6.2.1 Working principle and measuring instrument
  • 6.2.2 PPG operational configurations and signal characteristics
  • 6.2.3 Photoplethysmography challenges
  • 6.2.4 Noncontact photoplethysmography and potential physiological measurements
  • 6.3 Remote photoplethysmography methods
  • 6.3.1 Publicly available datasets
  • 6.3.2 Skin reflection model
  • 6.3.3 Architectural components of deep learning-based rPPG methods
  • 6.3.3.1 Region of interest selection
  • Face detection
  • Skin segmentation.
  • 6.3.3.2 Deep learning networks for rPPG signal extraction methods
  • 3D convolutional neural network-based rPPG methods
  • 2D CNN with sequence models or temporal shift module
  • DL architectures comprised of the combination of different convolutional variants based on spatial dimensions
  • Generative adversarial networks
  • Autoencoders
  • Transformers
  • Transformers with other NN architectures
  • Other methods
  • Summarizing architectures
  • 6.3.4 Signal processing techniques for physiological parameters estimation
  • 6.4 Limitations and recommendations for future research
  • 6.5 Conclusion
  • References
  • 7 Federated learning in healthcare applications
  • 7.1 Introduction
  • 7.2 Preliminaries
  • 7.2.1 Centralized and decentralized learning
  • 7.2.2 Federated learning
  • 7.2.3 FL training algorithms
  • 7.2.4 Impact of FL on stakeholders
  • 7.3 FL applications in healthcare
  • 7.3.1 Electronic health record mining
  • 7.3.2 Remote health monitoring
  • 7.3.3 Medical imaging
  • 7.3.4 Disease prediction
  • 7.4 Challenges and considerations
  • 7.4.1 Domain generalization
  • 7.4.2 Data and model heterogeneity
  • 7.4.3 Privacy and security
  • 7.4.4 System architecture and resource sharing
  • 7.4.5 Lack of dataset and training
  • 7.5 Conclusions and future scope
  • References
  • 8 Riemannian deep feature fusion with autoencoder for MEG depression classification in smart healthcare applications
  • 8.1 Introduction and motivation
  • 8.2 Literature review and state-of-the-art
  • 8.3 Problem definition
  • 8.4 Proposed solution
  • 8.4.1 Preprocessing
  • 8.4.2 Feature extraction
  • 8.4.2.1 Riemannian features
  • 8.4.2.2 Markov chain features
  • 8.4.3 Feature fusion
  • 8.4.4 Classification
  • 8.5 Experiment
  • 8.6 Conclusion and future work
  • References.
  • 9 Source localization of epileptiform MEG activity towards intelligent smart healthcare: a retrospective study
  • 9.1 Introduction and motivation
  • 9.2 Literature review and state-of-the-art
  • 9.3 MEG recordings
  • 9.4 MRI
  • 9.5 Problem definition
  • 9.6 Dataset description
  • 9.7 Proposed solution
  • 9.7.1 Data preprocessing
  • 9.7.2 Artifact detection and reduction
  • 9.7.3 MRI
  • 9.7.3.1 MRI data realigning, reslicing, and segmentation using SPM12
  • 9.7.3.2 MRI head model and source model
  • 9.7.4 Estimation of the epileptic foci
  • 9.8 Results and discussion
  • 9.9 Conclusion and future work
  • References
  • 10 Early classification of time series data: overview, challenges, and opportunities
  • 10.1 Introduction
  • 10.2 Overview
  • 10.2.1 Traditional classification of time series
  • 10.2.2 Early classification of time series
  • 10.3 Methods
  • 10.3.1 Instance-based methods
  • 10.3.2 Shapelet-based methods
  • 10.3.3 Model-based methods
  • 10.3.4 Other methods
  • 10.4 Data fusion
  • 10.5 Challenges
  • 10.5.1 Handling missing values
  • 10.5.2 Decision strategy
  • 10.5.3 Conflicting objectives
  • 10.5.4 Interpretability
  • 10.5.5 Data fusion
  • 10.6 Opportunities and future directions
  • 10.7 Conclusion
  • References
  • 11 Deep learning-based multimodal medical image fusion
  • 11.1 Introduction
  • 11.1.1 Fusion levels
  • 11.1.2 Preprocessing pipeline
  • 11.2 Literature survey and state-of-the-art
  • 11.2.1 Traditional techniques
  • 11.2.2 Deep learning-based techniques
  • 11.3 Proposed framework
  • 11.3.1 Feature extraction
  • 11.3.2 Feature fusion and reconstruction
  • 11.3.3 Implementation details
  • 11.4 Experimental results and discussion
  • 11.4.1 Evaluation setup
  • 11.4.2 Results and discussion
  • 11.4.2.1 Qualitative evaluation
  • 11.4.2.2 Quantitative evaluation
  • 11.5 Conclusion
  • References.
  • 12 Data fusion in Internet of Medical Things: towards trust management, security, and privacy
  • 12.1 Introduction
  • 12.2 Preliminaries of IoMT data fusion
  • 12.2.1 Characteristics of data in the context of IoMT
  • 12.2.2 IoMT data properties
  • 12.2.3 IoMT data fusion requirements and challenges
  • 12.2.3.1 IoMT data fusion requirements
  • 12.2.3.2 IoMT data fusion challanges
  • 12.2.4 Fusion metrics
  • 12.3 Traditional data fusion methods
  • 12.4 Data fusion in IoMT
  • 12.5 Privacy-enhanced data fusion
  • 12.5.1 Diverse duties of various layers of IoMT in data fusion
  • 12.5.2 Trust evaluation model for data fusion
  • 12.5.3 Source sensor selection
  • 12.6 Conclusion
  • Acknowledgment
  • References
  • 13 Feature fusion for medical data
  • 13.1 Introduction
  • 13.2 Morphological methods
  • 13.3 Component substitution-based methods
  • 13.3.1 How does PCA work?
  • 13.4 Multiscale decomposition-based methods
  • 13.4.1 Feature fusion with wavelet transform
  • 13.4.2 Wavelet-based combination feature fusion
  • 13.4.3 Feature fusion with contourlet methods
  • 13.4.4 Ridgelet transform
  • 13.4.5 Curvelet transform
  • 13.5 Deep learning methods for feature fusion
  • 13.6 Fuzzy logic
  • 13.7 Sparse representation methods
  • 13.8 Feature fusion on medical data
  • 13.9 Comparison of data fusion methods
  • 13.10 Future works
  • 13.11 Conclusion
  • References
  • 14 Review on hybrid feature selection and classification of microarray gene expression data
  • 14.1 Introduction and motivation
  • 14.2 Background, definitions, and notations
  • 14.2.1 Dataset description
  • 14.2.2 Feature selection and classification
  • 14.3 System definition
  • 14.3.1 Genetic algorithm (GA)
  • 14.3.2 Differential evolution (DE)
  • 14.3.3 Particle swarm optimization (PSO)
  • 14.3.4 Ant colony optimization (ACO)
  • 14.3.5 Tabu search (TS)
  • 14.3.6 Subset evaluation.