Ophthalmic Medical Image Analysis: 9th International Workshop, OMIA 2022, Held in Conjunction with MICCAI 2022, Singapore, Singapore, September 22, ... (Lecture Notes in Computer Science, 13576)

This book constitutes the refereed proceedings of the 9th International Workshop on Ophthalmic Medical Image Analysis, OMIA 2022, held in conjunction with the 25th International Conference on Medical Imaging and Computer-Assisted Intervention, MICCAI 2022, in Singapore, Singapore, in September 2022. The 20 papers presented at OMIA 2022 were carefully reviewed and selected from 33 submissions. The papers cover various topics in the field of ophthalmic medical image analysis and challenges in terms of reliability and validation, number and type of conditions considered, multi-modal analysis (e.g., fundus, optical coherence tomography, scanning laser ophthalmoscopy), novel imaging technologies, and the effective transfer of advanced computer vision and machine learning technologies. Preface Organization Contents AugPaste: One-Shot Anomaly Detection for Medical Images 1 Introduction 2 Methods 2.1 Construction of Lesion Bank 2.2 Synthesis of Anomalous Samples 2.3 Anomaly Detection Network 2.4 Implementation Details 3 Experiments and Results 3.1 Datasets 3.2 Evaluation Metric 3.3 Ablation Studies on EyeQ 3.4 Comparison with State-of-the-Art 4 Conclusion References Analysing Optical Coherence Tomography Angiography of Mid-Life Persons at Risk of Developing Alzheimer's Disease Later in Life 1 Introduction 2 Methodology 3 Results 3.1 Vessel Tortuosity Decreases in Risk Groups 3.2 Longitudinal Variations of Retinal Features in Risk Groups 4 Discussion 5 Conclusion References Feature Representation Learning for Robust Retinal Disease Detection from Optical Coherence Tomography Images 1 Introduction 2 Related Work 3 Methodology 3.1 Robust Feature Learning Architecture 3.2 Proposed Representation Learning Loss 3.3 Final Objective Function 4 Experiments 4.1 Data-Set Processing 4.2 Hyper-parameter Tuning 4.3 Performance Metrics 4.4 Quantitative Evaluation 4.5 Qualitative Evaluation 5 Conclusion and Future Work References GUNet: A GCN-CNN Hybrid Model for Retinal Vessel Segmentation by Learning Graphical Structures 1 Introduction 2 Method 2.1 GUNet 2.2 Graph Convolution 2.3 Graph Construction 3 Experiments 3.1 Datasets and Evaluation Metrics 3.2 Implementation Details 4 Results 4.1 Experiments on Fundus Photography 4.2 Experiments on SLO Images 4.3 Visualization 5 Conclusion References Detection of Diabetic Retinopathy Using Longitudinal 1 Introduction 2 Methods 2.1 Longitudinal Siamese 2.2 Longitudinal Self-supervised Learning 2.3 Longitudinal Neighbourhood Embedding 3 Dataset 4 Experiments and Results 4.1 Comparison of the Approaches on the Early Change Detection 4.2 Norm of Trajectory Vector Analyze 5 Discussion References Multimodal Information Fusion for Glaucoma and Diabetic Retinopathy Classification 1 Introduction 2 Methods 2.1 Early Fusion 2.2 Intermediate Fusion 2.3 Hierarchical Fusion 3 Material and Experiments 3.1 Data 3.2 Data Pre-processing 3.3 Implementation Details 4 Results 4.1 GAMMA Dataset 4.2 PlexEliteDR Dataset 5 Conclusion References Mapping the Ocular Surface from Monocular Videos with an Application to Dry Eye Disease Grading 1 Introduction 2 Proposed Method 3 Experiments and Results 3.1 SiGMoid 3.2 DED Diagnosis: Classification 4 Discussion and Conclusion References Rethinking Retinal Image Quality: Treating Quality Threshold as a Tunable Hyperparameter 1 Introduction 2 Methods 2.1 Quality Prediction on a Categorical Scale and Continuous Scale 2.2 Effect of Varying Image Quality Threshold 3 Experiments 3.1 Altering Quality Threshold on a Categorical Scale 3.2 Altering Quality Threshold on a Continuous Scale 3.3 Tuning on a Continuous Scale: Does it Confer Additional Value? 4 Discussions and Conclusions References Robust and Efficient Computation of Retinal Fractal Dimension Through Deep Approximation 1 Introduction 2 Deep Approximation of Retinal Traits (DART) 2.1 Motivation and Theory 2.2 Implementation 3 Experiments 3.1 Data 3.2 Results 4 Conclusion References Localizing Anatomical Landmarks in Ocular Images Using Zoom-In Attentive Networks 1 Introduction 2 Method 2.1 Zoom-In Module 2.2 Attentive Fusion Module 3 Experiments and Results 3.1 Datasets and Settings 3.2 Experimental Setup 3.3 Results and Discussion 4 Conclusions References Intra-operative OCT (iOCT) Super Resolution: A Two-Stage Methodology Leveraging High Quality Pre-operative OCT Scans 1 Introduction 2 Methods 2.1 Datasets 2.2 Two-Stage Super-Resolution Approach 2.3 Implementation Details 2.4 Evaluation Metrics 3 Results 3.1 Evaluation on Real iOCT Images 3.2 Evaluation on Pseudo iOCT Images 4 Discussion and Conclusions References Domain Adaptive Retinal Vessel Segmentation Guided by High-frequency Component 1 Introduction 2 Methodology 2.1 Fourier Domain Adaptation 2.2 High-frequency Component Extraction Based on Gaussian Filtering 2.3 Multi-input Deep Vessel Segmentation Model 3 Experiments 3.1 Experiment Settings 3.2 Comparison and Ablation Study 4 Conclusion References Tiny-Lesion Segmentation in OCT via Multi-scale Wavelet Enhanced Transformer 1 Introduction 2 Method 2.1 Overall Architecture 2.2 The Encoder Path 2.3 The Adaptive Multi-scale Transformer Module 3 Experiments and Results 3.1 Dataset 3.2 Implementation Detail 3.3 Comparison Study 3.4 Wavelet Feature Representation Visualization 3.5 Ablation Study 4 Conclusion References Dataset and Evaluation Algorithm Design for GOALS Challenge 1 Introduction 2 Dataset 3 Baseline 4 Evaluation 4.1 Task 1: Layer Segmentation 4.2 Task 2: Glaucoma Classification 5 Conclusion References Self-supervised Learning for Anomaly Detection in Fundus Image 1 Introduction 2 Methodology 2.1 Illumination Information Change Augmentation 2.2 Reconstruction for Reflectance Image 2.3 Semi-hard Negative Mining Strategy 3 Experiments and Result 3.1 Dataset 3.2 Anomaly Score 3.3 Ablation Study 3.4 Comparison with the State-of-the-Arts(SOTA) 3.5 Qualitative Analysis 4 Conclusion References GARDNet: Robust Multi-view Network for Glaucoma Classification in Color Fundus Images 1 Introduction 2 Related Works 3 Method 3.1 Preprocessing 3.2 Multi-view Classification Network 4 Datasets 5 Experimental Setup 6 Experiments and Results 7 Discussion 8 Conclusion References Fundus Photograph Defect Repair Algorithm Based on Portable Camera Empty Shot 1 Introduction 2 Related Work 2.1 Image Enhancement 2.2 Image Inpainting 3 The Proposed Method 3.1 Camera Empty Shot Image 3.2 Compensation Template 4 Evaluations 4.1 Data Set and Experimental Setup 4.2 Experimental Results 5 Conclusions References Template Mask Based Image Fusion Built-in Algorithm for Wide Field Fundus Cameras 1 Introduction 2 Related Work 3 The Proposed Method 3.1 Wide Field Fundus Image Pre-processing 3.2 Color and Brightness Normalization Based on Poisson Fusion 3.3 Image Fusion Based on Template Mask 3.4 Adaptive Brightness Adjustment 4 Evaluations 4.1 Data Set and Experimental Setup 4.2 Results and Analysis 5 Conclusions References Investigating the Vulnerability of Federated Learning-Based Diabetic Retinopathy Grade Classification to Gradient Inversion Attacks 1 Introduction 2 Methods and Materials 2.1 Data 2.2 Federated Learning and Gradient Inversion Attack Framework 2.3 Segmentation Matching Score 2.4 Evaluation 3 Results 3.1 Gradient Inversion Attack Performance 3.2 Extracting Identifiable Clinical Features from Reconstructed Images 4 Conclusions References Extraction of Eye Redness for Standardized Ocular Surface Photography 1 Introduction 2 Methods 2.1 High-Resolution, Standardized Ocular Surface Photography System 2.2 Image Data Set 2.3 Automated Sclera Detection 2.4 Eye Redness Extraction 2.5 Ocular Surface Tile Annotation 3 Experiments and Results 3.1 Ocular Surface Tile Annotation 3.2 Automated Tile Classification 3.3 Redness Extraction 4 Conclusion References Author Index