Myopic Maculopathy Analysis: MICCAI Challenge MMAC 2023, Held in Conjunction with MICCAI 2023, Virtual Event, October 8–12, 2023, Proceedings (Lecture Notes in Computer Science Book 14563)

This book constitutes the MICCAI Challenge, MMAC 2023, that held in Conjunction with MICCAI 2023, Vancouver, BC, Canada, which took place in October 2023. The 11 long papers included in this volume presents a wide range of state-of-the-art deep learning methods developed for the various tasks presented in the challenge. Preface Organization Contents Automated Detection of Myopic Maculopathy in MMAC 2023: Achievements in Classification, Segmentation, and Spherical Equivalent Prediction 1 Introduction 2 Materials and Methods 2.1 Datasets 2.2 Task 1: Classification of Myopic Maculopathy 2.3 Task 2: Segmentation of Myopic Maculopathy Plus Lesions 2.4 Task 3: Prediction of Spherical Equivalent 2.5 Implementation Details 3 Results 3.1 Task 1: Classification of Myopic Maculopathy 3.2 Task 2: Segmentation of Myopic Maculopathy Plus Lesions 3.3 Task 3: Prediction of Spherical Equivalent 4 Discussion and Conclusions References Swin-MMC: Swin-Based Model for Myopic Maculopathy Classification in Fundus Images 1 Introduction 2 Method 2.1 Enhanced ArcFace Loss with 3 Sub-centers 2.2 Weak Label 3 Experiments 3.1 Dataset and Evaluation Measures 3.2 Image Preprocessing and Augmentation 3.3 Implementation Details 4 Results and Discussion 4.1 Results on Testing Set 4.2 Visualization Heatmap Analysis 4.3 Ablation Study in Further Test Phase 4.4 Limitation and Future Work 5 Conclusion References Towards Label-Efficient Deep Learning for Myopic Maculopathy Classification 1 Introduction 2 Related Work 3 Methodology 3.1 Pre-training 3.2 Network Architecture 3.3 Pseudo Labeling 3.4 Image Resolution 3.5 Loss Function 3.6 Model Ensemble 4 Experiment 4.1 Dataset 4.2 Implementation Details 4.3 Evaluation Metrics 4.4 Results on the Validation Set 4.5 Results on the MMAC Leaderboard 5 Conclusion References Ensemble Deep Learning Approaches for Myopic Maculopathy Plus Lesions Segmentation 1 Introduction 2 Related Work 3 Methodology 3.1 Network Architecture 3.2 Loss Function 3.3 Model Ensemble 4 Experiments 4.1 Dataset 4.2 Implementation Details 4.3 Evaluation Metrics 4.4 Results on the Validation Set 4.5 Results on the Leaderboard 4.6 Visual Segmentation Results 5 Conclusion References Beyond MobileNet: An Improved MobileNet for Retinal Diseases 1 Introduction 2 Related Work 2.1 CNN-Based Method for RD Diagnosis 2.2 VIT-Based Method for RD Diagnosis 3 Methods 3.1 Network Design 3.2 Training Techniques 4 MMAC - Classification of Myopic Maculopathy 4.1 Dataset and Evaluation Metrics 4.2 Implementation Details 4.3 Experimental Results 5 Conclusion References Prediction of Spherical Equivalent with Vanilla ResNet 1 Introduction 2 Related Works 3 Methods 4 Results 5 Discussion: The Significance of Proper Data Augmentation 6 Conclusion References Semi-supervised Learning for Myopic Maculopathy Analysis 1 Introduction 2 Related Work 3 Datasets 4 Segmentation of Myopic Maculopathy Plus Lesions 5 Prediction of Spherical Equivalent 6 Conclusions References A Clinically Guided Approach for Training Deep Neural Networks for Myopic Maculopathy Classification 1 Introduction 2 Methods 2.1 Datasets and Pre-processing 2.2 Image Synthesis Pipeline Guided by Clinical Domain Knowledge 2.3 Mix-Up Augmentation 2.4 Evaluation Metrics 2.5 Training Details 2.6 Ensemble Prediction via Test-Time Augmentation 3 Results 4 Conclusions and Future Directions References Classification of Myopic Maculopathy Images with Self-supervised Driven Multiple Instance Learning Network 1 Introduction 2 Related Work 2.1 Deep Learning in Myopic Maculopathy Analysis 2.2 Multiple Instance Learning 2.3 Self-supervised Learning 3 Methodology 3.1 Generative Data Augmentation 3.2 Backbone Architecture 4 Experiments 4.1 Datasets and Implementation 4.2 Results 5 Conclusion References Self-supervised Learning and Data Diversity Based Prediction of Spherical Equivalent 1 Introduction 2 Our Solution 2.1 Baseline 2.2 Self-supervised Learning 2.3 Increase Data Diversity 2.4 Part of Data 2.5 Test-Time Augmentation 3 Experiment 3.1 Implement Details 3.2 Experiment Results 4 Conclusion References Myopic Maculopathy Analysis Using Multi-task Learning and Pseudo Labeling 1 Introduction 2 Related Work 3 Method 3.1 Multi-task Learning 3.2 Pseudo-labeling 4 Results 5 Conclusion References Author Index