Towards the Automatization of Cranial Implant Design in Cranioplasty II: Second Challenge, AutoImplant 2021, Held in Conjunction with MICCAI 2021, ...

Description

This book constitutes the Second Automatization of Cranial Implant Design in Cranioplasty Challenge, AutoImplant 2021, which was held in conjunction with the 24th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2021, in Strasbourg, France, in September, 2021. The challenge took place virtually due to the COVID-19 pandemic. The 7 papers are presented together with one invited paper, one qualitative evaluation criteria from neurosurgeons and a dataset descriptor. This challenge aims to provide more affordable, faster, and more patient-friendly solutions to the design and manufacturing of medical implants, including cranial implants, which is needed in order to repair a defective skull from a brain tumor surgery or trauma. The presented solutions can serve as a good benchmark for future publications regarding 3D volumetric shape learning and cranial implant design. Preface Organization Contents Personalized Calvarial Reconstruction in Neurosurgery 1 Indications for Craniectomy 2 Indications for Cranioplasty 3 Personalized Calvarial Reconstruction 4 Challenges and Requirements 5 Perspective References Qualitative Criteria for Feasible Cranial Implant Designs 1 Introduction 2 Methods 2.1 Data 2.2 Evaluation 2.3 3D Printing 3 Results 3.1 Qualitative Analysis 3.2 Quantitative Analysis 3.3 Evaluation of 3D Printed Implant Designs 4 Discussion 5 Conclusion References Segmentation of Defective Skulls from CT Data for Tissue Modelling 1 Introduction 2 Proposed Method 3 Experiments 3.1 Dataset 3.2 Metrics 3.3 Experimental Design and Results 4 Discussion 5 Conclusions References Improving the Automatic Cranial Implant Design in Cranioplasty by Linking Different Datasets 1 Introduction 2 Dataset 3 Methods 3.1 Overview 3.2 Preprocessing 3.3 Dataset Linking and Augmentation 3.4 U-Net-Based Segmentation 3.5 Postprocessing 4 Results 4.1 Overview 4.2 Task 1 4.3 Task 2 4.4 Task 3 5 Discussion and Conclusion References Learning to Rearrange Voxels in Binary Segmentation Masks for Smooth Manifold Triangulation 1 Introduction 1.1 Background 1.2 Related Work 2 Dataset 3 Method 3.1 Learning Voxel Rearrangement 3.2 Hierarchical Image Synthesis 4 Experiment and Results 4.1 Interpolation, Patch-Wise Skull Shape Completion and Voxel Rearrangement 4.2 Image Synthesis 5 Discussion and Future Work 6 Conclusion References A U-Net Based System for Cranial Implant Design with Pre-processing and Learned Implant Filtering 1 Introduction 2 Datasets 2.1 SkullFix 2.2 SkullBreak 2.3 Challenge Test Datasets 3 Methods 3.1 Preprocessing 3.2 U-Net 3.3 Post-processing 3.4 Analysis 4 Experiments and Results 4.1 Comparison of Different Training Datasets 4.2 Comparison of Cropping Methods 4.3 Comparison of Implant Filtering Methods 4.4 Performance on AutoImplant Challenge 5 Discussion and Future Work References Sparse Convolutional Neural Network for Skull Reconstruction 1 Introduction 2 Minkowski Engine 2.1 Sparse Convolutions 3 Methods 3.1 Data Preprocessing and Postprocessing 3.2 Setup and Network Design 3.3 Training Configuration 3.4 Testing Configuration 4 Results 4.1 Evaluation Metrics 4.2 Memory Consumption 4.3 Implant Generation Issues 5 Conclusion References Cranial Implant Prediction by Learning an Ensemble of Slice-Based Skull Completion Networks 1 Introduction 2 Methodology 2.1 Dataset 2.2 Motivation 2.3 Architecture 3 Experimentation and Results 4 Conclusions References PCA-Skull: 3D Skull Shape Modelling Using Principal Component Analysis 1 Introduction 2 Method 2.1 Registration 2.2 PCA-Averager 2.3 Inverse-Registration 2.4 Post-processing 3 Experiment and Results 3.1 Dataset 3.2 Results on Task 2 of the AutoImplant 2021 Challenge 3.3 Results on Task 3 of the AutoImplant 2021 Challenge 3.4 Visualization of the Principle Components 4 Discussion and Future Work 5 Conclusion References Cranial Implant Design Using V-Net Based Region of Interest Reconstruction 1 Introduction 2 Proposed Method 2.1 Defect Localization 2.2 Defective Region Extraction 2.3 Implant Reconstruction 2.4 V-Net Architecture 3 Results 3.1 Dataset 3.2 Training 3.3 Evaluations 3.4 Ablation Study 3.5 Manufacturing Cranial Implant Through 3D Printing 4 Conclusions and Discussions References Author Index