Imaging systems for GI endoscopy, and graphs in biomedical image analysis : first MICCAI workshop, ISGIE 2022, and fourth MICCAI Workshop, GRAIL 2022, held in conjunction with MICCAI 2022, Singapore, September 18, 2022 : proceedings

"This book constitutes the refereed proceedings of the first MICCAI Workshop, ISGIE 2022, Imaging Systems for GI Endoscopy, and the Fourth MICCAI Workshop, GRAIL 2022, GRaphs in biomedicAL Image and analysis, held in conjunction with MICCAI 2022, Singapore, September 18, 2022. ISGIE 2022 accepted 6 papers from the 8 submissions received.This workshop focuses on novel scientific contributions to vision systems, imaging algorithms as well as the autonomous system for endorobot for GI endoscopy. This includes lesion and lumen detection, as well as 3D reconstruction of the GI tract and hand-eye coordination. GRAIL 2022 accepted 6 papers from the 10 submissions received. The workshop aims to bring together scientists that use and develop graph-based models for the analysis of biomedical images and to encourage the exploration of graph-based models for difficult clinical problems within a variety of biomedical imaging contexts." ISGIE Preface Preface GRAIL 2022 Organization Contents Imaging Systems for GI Endoscopy Light Adaptation for Classification of the Upper Gastrointestinal Sites 1 Introduction 2 Methods 2.1 Motivation 2.2 Common Classification Model 2.3 Light Classification Head 2.4 Reconstruction Decoder 2.5 Implementation Details 3 Experiments 3.1 Dataset 3.2 Experiment Details 3.3 Results 4 Conclusion References Criss-Cross Attention Based Multi-level Fusion Network for Gastric Intestinal Metaplasia Segmentation 1 Introduction 2 Related Work 3 Method 3.1 Criss-cross Attention Based Feature Fusion Encoder 3.2 Feature Activation Map Guided Multi-level Decoder 4 Experimental Results 4.1 Dataset 4.2 Ablation Study 4.3 Quantitative Results 4.4 Qualitative Results 5 Conclusions References Colonoscopy Landmark Detection Using Vision Transformers 1 Introduction 2 Related Work 2.1 Landmark Detection 2.2 Visual Feature Backbones and Optimizers 3 Data Collection 3.1 Annotations and Cross-Validation 3.2 Snapshots Dataset 3.3 Training Dataset 4 Problem Definition 5 Architecture 6 Training Pipeline 7 Results 8 Inference and Future Work References Real-Time Lumen Detection for Autonomous Colonoscopy 1 Introduction 2 Methods 3 Experimental Set-Up, Results and Discussion 4 Conclusions References SuperPoint Features in Endoscopy 1 Introduction 2 Related Work 3 SuperPoint in Endoscopy 3.1 SuperPoint Models Considered 3.2 SuperPoint Matching 4 Experiments 5 Conclusions References Estimating the Coverage in 3D Reconstructions of the Colon from Colonoscopy Videos 1 Introduction 2 Related Works 3 Coverage Estimation of 3D Colon Reconstructions 3.1 Dataset 3.2 Method 4 Results 5 Conclusion References Graphs in Biomedical Image Analysis Modular Graph Encoding and Hierarchical Readout for Functional Brain Network Based eMCI Diagnosis 1 Introduction 2 Method 2.1 Brain Functional Network Construction 2.2 Graph Convolutional Network 2.3 Topology-Focused Brain Encoder 2.4 Hierarchical Brain Readout 3 Experiments 3.1 Dataset 3.2 Implementation 3.3 Comparison of Methods 4 Conclusion References Bayesian Filtered Generation of Post-surgical Brain Connectomes on Tumor Patients 1 Introduction 2 Methods 3 Results 4 Discussion References Deep Cross-Modality and Resolution Graph Integration for Universal Brain Connectivity Mapping and Augmentation 1 Introduction 2 Proposed Method 3 Results and Discussion 4 Conclusion References Using Hierarchically Connected Nodes and Multiple GNN Message Passing Steps to Increase the Contextual Information in Cell-Graph Classification 1 Introduction 2 Method 2.1 Data 2.2 Cell-Graph Formation 2.3 Supernodes 2.4 Model Architecture 2.5 Experiential Evaluation 3 Results 3.1 The Inclusion of Supernodes 3.2 Multiple GNN Message Passing Steps 4 Discussion 5 Conclusion References TaG-Net: Topology-Aware Graph Network for Vessel Labeling 1 Introduction 2 Method 2.1 Vessel Centerline Extraction 2.2 Topology-Aware Graph Construction 2.3 Topology-Aware Graph Network for Centerline Labeling 2.4 Vessel Mask Labeling 3 Experiments 4 Conclusion References Transforming Connectomes to ``Any'' Parcellation via Graph Matching 1 Introduction 2 Connectome-to-Connectome (C2C) Mapping 2.1 Overview of C2C 2.2 Spectral Embedding (Algorithm 2) 2.3 Gromov-Wasserstein Discrepancy 2.4 Node-Based Mapping via Optimal Transport (Algorithm 4) 2.5 Connetomes Estimation 3 Results 3.1 Datasets 3.2 Graph Matching Methods 3.3 Evaluating C2C on HCP-D Structural Connectomes 3.4 Age Prediction Using Estimated Connectomes 4 Discussion and Conclusions References Author Index