Head and Neck Tumor Segmentation and Outcome Prediction: Third Challenge, HECKTOR 2022, Held in Conjunction with MICCAI 2022, Singapore, September 22, ... (Lecture Notes in Computer Science)

Preface
Organization
Contents
Overview of the HECKTOR Challenge at MICCAI 2022: Automatic Head and Neck Tumor Segmentation and Outcome Prediction in PET/CT
	1 Introduction: Research Context
	2 Dataset
		2.1 Mission of the Challenge
		2.2 Challenge Dataset
	3 Task 1: Segmentation
		3.1 Methods: Reporting of Challenge Design
		3.2 Results: Reporting of Segmentation Task Outcome
	4 Task 2: Outcome Prediction
		4.1 Methods: Reporting of Challenge Design
		4.2 Results: Reporting of Challenge Outcome
	5 Discussion: Putting the Results into Context
		5.1 Outcomes and Findings
		5.2 Limitations of the Challenge
	6 Conclusions
	Appendix 1:  Challenge Information
	Appendix 2:  Image Acquisition Details
	References
Automated Head and Neck Tumor Segmentation from 3D PET/CT HECKTOR 2022 Challenge Report
	1 Introduction
	2 Materials and Methods
		2.1 Training Method
		2.2 Optimization
	3 Results
	4 Conclusion
	References
A Coarse-to-Fine Ensembling Framework for Head and Neck Tumor and Lymph Segmentation in CT and PET Images
	1 Introduction
	2 Method
		2.1 Dataset
		2.2 Network Architecture
		2.3 Training Details
	3 Results and Discussion
	References
A General Web-Based Platform for Automatic Delineation of Head and Neck Gross Tumor Volumes in PET/CT Images
	1 Introduction
	2 Materials and Methods
		2.1 Network Architecture
		2.2 Data Preprocessing
		2.3 Training Procedure
		2.4 Post Processing Steps
	3 Dedicated Workflow Pipeline for Practical Clinical Usage
	4 Results and Discussion
	5 Conclusions
	References
Octree Boundary Transfiner: Efficient Transformers for Tumor Segmentation Refinement
	1 Introduction
		1.1 Related Work
	2 Dataset
		2.1 Data Preprocessing
		2.2 Data Augmentation
	3 Methods
		3.1 Losses
		3.2 Incoherence Maps
		3.3 Patch Selection
		3.4 Training Details
	4 Results
	5 Conclusion
	References
Head and Neck Primary Tumor and Lymph Node Auto-segmentation for PET/CT Scans
	1 Introduction
	2 Materials and Methods
		2.1 Dataset
		2.2 Preprocessing
		2.3 Augmentations
		2.4 Architecture Overview
		2.5 nnU-Net
		2.6 MNet
		2.7 Swin Transformer
	3 Results
	4 Discussion
	5 Conclusion
	References
Fusion-Based Automated Segmentation in Head and Neck Cancer via Advance Deep Learning Techniques
	1 Introduction
	2 Materials and Methods
		2.1 Dataset, PET/CT Acquisition
		2.2 Proposed Deep Learning Algorithm
		2.3 Analysis Procedure
	3 Results and Discussion
	4 Conclusions
	References
Stacking Feature Maps of Multi-scaled Medical Images in U-Net for 3D Head and Neck Tumor Segmentation
	1 Introduction
	2 Related Work
	3 Methods
		3.1 Preprocessing
		3.2 Stacked Multi-scale `U\' Shape Network
		3.3 Optimization and Data Augmentation
	4 Result
		4.1 HECKTOR 2022 Datasets
		4.2 Implementation Details
		4.3 Hector 2022 Test Result
		4.4 Qualitative Results
	5 Discussion
	References
A Fine-Tuned 3D U-Net for Primary Tumor and Affected Lymph Nodes Segmentation in Fused Multimodal Images of Oropharyngeal Cancer
	1 Introduction
	2 Methods
		2.1 Study Design
		2.2 Image Preprocessing
		2.3 Deep Learning Model
	3 Results
	4 Discussion
	5 Conclusion
	References
A U-Net Convolutional Neural Network with Multiclass Dice Loss for Automated Segmentation of Tumors and Lymph Nodes from Head and Neck Cancer PET/CT Images
	1 Introduction
	2 Materials and Methods
		2.1 Data Description
		2.2 Image Preprocessing
		2.3 Network Architecture
		2.4 Model Training
		2.5 Image Postprocessing
	3 Results
		3.1 Segmentation Performance
		3.2 Model Performance Analysis on Slices Containing GTVp or GTVn
	4 Conclusion and Discussion
	References
Multi-scale Fusion Methodologies for Head and Neck Tumor Segmentation
	1 Introduction
	2 Method
		2.1 Submission 1: OARFocalFuseNet
		2.2 Submission 2: 3D Multi-scale Fusion Network
		2.3 Submission 3: SwinUNETR
	3 Experiments
		3.1 Data Pre-processing and Data Augmentation
		3.2 Training Details
	4 Results and Discussion
	5 Conclusion
	References
Swin UNETR for Tumor and Lymph Node Segmentation Using 3D PET/CT Imaging: A Transfer Learning Approach
	1 Introduction
	2 Methods and Materials
		2.1 Data
		2.2 Data Preprocessing
		2.3 Model Architecture
		2.4 Experiments
		2.5 Quantitative Evaluation
	3 Results
	4 Discussion and Conclusion
	References
Simplicity Is All You Need: Out-of-the-Box nnUNet Followed by Binary-Weighted Radiomic Model for Segmentation and Outcome Prediction in Head and Neck PET/CT
	1 Introduction
	2 Materials and Methods
		2.1 Data
		2.2 Tumor and Lymph Node Segmentation
		2.3 Outcome Prediction
	3 Results
		3.1 Segmentation Evaluation
		3.2 Qualitative Assessment
		3.3 Performance of the Outcome Prediction Model
		3.4 Resilience to the Curse of Dimensionality
		3.5 Feature Importance
	4 Discussion
		4.1 Segmentation
		4.2 Binary-weighted Model
	5 Conclusions
	References
Radiomics-Enhanced Deep Multi-task Learning for Outcome Prediction in Head and Neck Cancer
	1 Introduction
	2 Materials and Methods
		2.1 Patients
		2.2 Radiomics-Enhanced Deep Multi-task Framework
		2.3 Deep Multi-task Survival Model (DeepMTS)
		2.4 Automatic Radiomics
		2.5 Image Preprocessing
		2.6 Training and Inference
		2.7 Ensemble
		2.8 Evaluation Metrics
	3 Results and Discussion
		3.1 Outcome Prediction
		3.2 Tumor Segmentation
	4 Conclusion and Limitations
	References
Recurrence-Free Survival Prediction Under the Guidance of Automatic Gross Tumor Volume Segmentation for Head and Neck Cancers
	1 Introduction
	2 Material and Methods
		2.1 Head and NeCK TumOR 2022 (HECKTOR 2022) Dataset
		2.2 Overall Architecture
		2.3 Gross Tumor Volume Segmentation (Task 1)
		2.4 Recurrence-Free Survival Prediction (Task 2)
	3 Results
		3.1 Gross Tumor Volume Segmentation (Task 1)
		3.2 Recurrence-Free Survival Prediction (Task 2)
	4 Discussion and Conclusion
	References
Joint nnU-Net and Radiomics Approaches for Segmentation and Prognosis of Head and Neck Cancers with PET/CT Images
	1 Introduction
	2 Method
		2.1 Dataset Description
		2.2 Pre-processing
		2.3 Task 1: Segmentation Prediction
		2.4 Task 2: Prognosis Prediction
		2.5 Evaluation Metrics
	3 Results
		3.1 Task 1: Segmentation Results
		3.2 Task 2: Prognostic Prediction Results
	4 Discussion
	5 Conclusion
	References
MLC at HECKTOR 2022: The Effect and Importance of Training Data When Analyzing Cases of Head and Neck Tumors Using Machine Learning
	1 Introduction
	2 Methods
		2.1 Task 1: Segmentation of CT and PET Scans
		2.2 Task 2: Prediction of Recurrence-Free Survival
	3 Discussion and Results
		3.1 Task 1
		3.2 Task 2
	4 Conclusion
	References
Towards Tumour Graph Learning for Survival Prediction in Head & Neck Cancer Patients
	1 Introduction
	2 Materials and Methods
		2.1 Localization
		2.2 Segmentation
		2.3 Survival Prediction
	3 Experimental Set up
		3.1 Data Splitting
		3.2 Data Preprocessing and Augmentations
		3.3 Implementation Details
	4 Results
		4.1 Segmentation Results
		4.2 Survival Prediction Results
	5 Discussion and Conclusion
	References
Combining nnUNet and AutoML for Automatic Head and Neck Tumor Segmentation and Recurrence-Free Survival Prediction in PET/CT Images
	1 Introduction
	2 Method
		2.1 Preprocessing
		2.2 Proposed Method
		2.3 Post-processing
	3 Experiments
		3.1 Dataset and Evaluation Measures
		3.2 Implementation Details
	4 Results and Discussion
		4.1 Quantitative Results on Validation Set
		4.2 Qualitative Results on Validation Set for Task 1
	5 Conclusion
	References
Head and Neck Cancer Localization with Retina Unet for Automated Segmentation and Time-To-Event Prognosis from PET/CT Images
	1 Introduction
	2 Methods
		2.1 Dataset
		2.2 Tumor Localization and Image Preprocessing
		2.3 Auto-segmentation
		2.4 RFS Prognosis
	3 Results
		3.1 Tumor Localization with Retina Unet
		3.2 Auto-segmentation Task
		3.3 Prognosis Task
	4 Conclusion
	References
HNT-AI: An Automatic Segmentation Framework for Head and Neck Primary Tumors and Lymph Nodes in FDG- PET/CT Images
	1 Introduction
	2 Material and Methods
		2.1 Data
		2.2 Data Preprocessing
		2.3 Network Architecture
		2.4 Experimental Settings
		2.5 Loss Function
		2.6 Post Processing
		2.7 Evaluation Metric
	3 Results
		3.1 Five-Fold Cross-Validation
		3.2 Test Set
		3.3 Tumor and Lymph Node Volume as a Biomarker for Prognosis
	4 Discussion
	5 Conclusion
	References
Head and Neck Tumor Segmentation with 3D UNet and Survival Prediction with Multiple Instance Neural Network
	1 Introduction
	2 Methods
		2.1 Image Pre-processing
		2.2 Segmentation Model
		2.3 Image Post-processing
		2.4 Feature Extraction
		2.5 Prediction Model
	3 Experimental Results
		3.1 Implementation Details
		3.2 Primary Tumor and Lymph Node Metastasis Segmentation
		3.3 Recurrence-Free Survival Prediction
	4 Discussion and Conclusion
	References
Deep Learning and Machine Learning Techniques for Automated PET/CT Segmentation and Survival Prediction in Head and Neck Cancer
	1 Introduction
	2 Material and Methods
		2.1 Patient Data
		2.2 Feature Extraction Using Autoencoder Algorithm
		2.3 AI Algorithms
		2.4 Analysis Procedure
	3 Results and Discussion
	4 Conclusions
	References
Deep Learning and Radiomics Based PET/CT Image Feature Extraction from Auto Segmented Tumor Volumes for Recurrence-Free Survival Prediction in Oropharyngeal Cancer Patients
	1 Introduction
	2 Method
		2.1 Data and Preprocessing
		2.2 Radiomics Model
		2.3 Deep Learning Model
		2.4 Model Evaluation
	3 Results
		3.1 General Data Description
		3.2 Radiomics Model
		3.3 Deep Learning Model
	4 Discussion
	5 Conclusion
	References
Author Index