This book covers the state-of-the-art techniques of fundus imaging for the diagnosis of retinal diseases. It is part of a three-volume work that describes the latest imaging techniques in which to bring optical coherence tomography (OCT), fundus Imaging and optical coherence tomography angiography (OCTA) to accurately facilitate the diagnosis of retinal diseases. Clinical disorders of the retina have been attracting the attention of researchers, aiming at reducing the blindness rate. This includes uveitis, diabetic retinopathy, macular edema, endophthalmitis, proliferative retinopathy, age-related macular degeneration and glaucoma. Treatment is significantly dependent on having early and accurate diagnosis, which can be significantly improved by employing the techniques described in the book.
Key features
• Provides a comprehensive overview of all pertinent topics related to fundus imaging techniques, applicable to diagnosis of eye disorders.
• Offers a unique coverage of Neural Networks in distinguishing eye diseases.
• Machine learning techniques are presented in detail throughout.
• Many of the chapter contributors are world-class researchers.
• Extensive references will be provided at the end of each chapter to enhance further study.
Author(s): Aymal El-Baz, Jasjit S. Suri
Publisher: IOP Publishing
Year: 2022
Language: English
Pages: 237
City: Bristol
PRELIMS.pdf
Preface
Acknowledgements
Editor biographies
Ayman El-Baz
Jasjit S Suri
List of contributors
CH001.pdf
Chapter 1 Texture interpretability of fundus imaging and diabetic retinopathy: a review
1.1 Introduction
1.2 DR interpretability approaches
1.2.1 Attention based models
1.2.2 Feed-forward style localization maps
1.2.3 Occlusion based models
1.2.4 Hand-crafted texture analysis methods
1.2.5 Other approaches
1.3 Texture challenges in a fundus imaging dataset
1.3.1 Intensity based analysis is challenging for DR classification
1.3.2 Image gradient analysis is challenging for DR classification
1.4 Detecting texture patterns with the aid of deep learning
1.4.1 Convolutional neural network model
1.4.2 Kernel clustering
1.4.3 Visualization
1.5 Limitations and challenges
1.6 Conclusion
References
CH002.pdf
Chapter 2 A two-phase novel optic disc detection algorithm based on vesselness distribution and a fuzzy classifier using vessel ramification and fundus color features
2.1 Introduction
2.2 Materials and methods
2.2.1 Materials
2.2.2 Methods
2.3 Evaluation and results
2.4 Discussion and conclusion
Acknowledgments
References
CH003.pdf
Chapter 3 Application of enface image registration/alignment to introduce new ocular imaging biomarkers
3.1 Introduction
3.1.1 Anatomic landmarks of the eye
3.1.2 Enface image modalities
3.2 Feature extraction
3.2.1 Vessel segmentation
3.2.2 Optic disc segmentation
3.2.3 FAZ segmentation
3.3 Registration of fundus and OCT images
3.4 Registration and alignment of ocular images in the right and left eyes
3.4.1 Symmetry analysis of vessel maps in the right and left eyes
3.4.2 Symmetry analysis of macular OCTs in the right and left eyes
3.4.3 Alignment of ONH OCTs in the right and left eyes
3.5 Registration of FFA and SLO images
3.5.1 FFA and OCT registration using SLO images
3.5.2 MA detection in OCT and FFA
3.6 Conclusion
References
CH004.pdf
Chapter 4 Existing techniques used for retinal image analysis in the automated detection and prediction of AMD
4.1 Introduction
4.2 Existing automated diagnosis methods used for AMD
4.2.1 Pre-processing techniques
4.2.2 Retinal component extraction from fundoscopic images
4.3 Retinal component extraction from OCT images
4.3.1 Retinal layer extraction
4.3.2 Drusen detection from extracted retinal layers
4.3.3 AMD lesion detection
4.4 Comparative analysis of diagnosis systems
4.5 Research limitations and challenges
4.6 Conclusion
References
CH005.pdf
Chapter 5 Interobserver variability in the determination of diabetic retinopathy and quality of fundus image
5.1 Introduction
5.2 Fundus imaging
5.2.1 The diabetic retinopathy classification process
5.2.2 Fundus image quality
5.3 Interobserver variability and reliability of DR classification
5.4 Image quality
5.5 Implications for clinical practice
5.6 Conclusion
References
CH006.pdf
Chapter 6 Computer-aided diagnosis of Plus disease in preterm infants
6.1 Introduction
6.1.1 Retinal image analysis
6.2 Computer-aided diagnosis system
6.2.1 Resources and evaluation metrics
6.3 Development of a computer-aided diagnostic system
6.3.1 Tortuosity detection and quantification using a fully convolutional neural network
6.4 Results and discussion
6.5 Conclusions
References
CH007.pdf
Chapter 7 Retinal disease management using fundus autofluorescence images
7.1 Introduction
7.2 Materials and methods
7.2.1 Fundus autoflourescence image acquisition
7.2.2 ETDRS-based statistics as the feature set
7.2.3 Support vector machine (SVM) classifier
7.2.4 Performance evaluation
7.3 Experimental results
7.4 Additional performance comparison
7.5 Possible progress monitoring
7.6 Discussion and future course
Appendix A Theoretical connection
Acknowledgments
References
CH008.pdf
Chapter 8 A review of mainstream ophthalmic AI algorithms: advances, limitations, and challenges
8.1 Brief introduction
8.2 Classification/detection
8.2.1 Diabetic retinopathy classification/grading
8.2.2 Age-related macular degeneration
8.2.3 Other common retinal diseases
8.2.4 The challenge of multiple retinal disease recognition
8.3 Segmentation
8.3.1 Lesion segmentation
8.3.2 Vessel segmentation
8.3.3 Optic disc/cup segmentation
References
CH009.pdf
Chapter 9 Diabetic retinopathy detection and classification through fundus images using AlexNet
9.1 Introduction
9.2 Diabetic retinopathy detection and classification
9.2.1 Diabetic retinopathy
9.2.2 Convolutional neural network
9.2.3 Method
9.3 Results
9.3.1 AlexNet architecture
9.3.2 The result
9.4 Conclusion
References
CH010.pdf
Chapter 10 A framework for joint cup and disc segmentation in fundus images
10.1 Introduction
10.2 Glaucoma detection using REFUGE data
10.2.1 Preprocessing REFUGE2 data
10.2.2 Relation between fundus image and polar transformed image
10.2.3 Implementation
10.3 Conclusion
10.3.1 Limitations of the research
References
