Diabetic Foot Ulcers Grand Challenge: Third Challenge, DFUC 2022, Held in Conjunction with MICCAI 2022, Singapore, September 22, 2022, Proceedings

Preface
Organization
Contents
Quantifying the Effect of Image Similarity on Diabetic Foot Ulcer Classification
	1 Introduction
	2 Related Work
	3 Methodology
		3.1 Dataset Description
		3.2 Fuzzy Algorithm
		3.3 Identification of Similar Images
		3.4 Removal of Similar Images as Determined by Similarity Thresholds
	4 Image Similarity Analysis
		4.1 Train Set Image Similarity
		4.2 Test Set Image Similarity
		4.3 Train and Test Set Image Similarity
		4.4 Inter-class Image Similarity
		4.5 Model Training
	5 Results and Discussion
		5.1 Baseline Results
		5.2 Results on the Curated Datasets
	6 Conclusion
	References
DFUC2022 Challenge Papers
HarDNet-DFUS: Enhancing Backbone and Decoder of HarDNet-MSEG for Diabetic Foot Ulcer Image Segmentation
	1 Introduction
	2 Method
		2.1 HarDNetV2 – Channel Balanced HarDNet
		2.2 Decoder
		2.3 Model Ensemble
		2.4 Loss Function
		2.5 Post-processing
	3 Experiments
		3.1 Settings
		3.2 Dataset
		3.3 Experiment Results
	4 Conclusion and Future Work
	References
OCRNet for Diabetic Foot Ulcer Segmentation Combined with Edge Loss
	1 Introduction
	2 Methodology
		2.1 Datasets
		2.2 Network
		2.3 Training
		2.4 Post-processing
		2.5 Evaluation Metrics
	3 Experiments
	4 Conclusions
	References
On the Optimal Combination of Cross-Entropy and Soft Dice Losses for Lesion Segmentation with Out-of-Distribution Robustness
	1 Introduction
	2 Methodology
		2.1 The Cross-Entropy Loss, the Dice Loss, and Their Combinations
		2.2 On the Definition of In-Distribution and Out-of-Distribution Data for Binary Segmentation Problems
		2.3 Model Training Details
	3 Experimental Results
		3.1 Data and Performance Measures
		3.2 Performance Evaluation and Discussion
	4 Conclusion
	References
Capture the Devil in the Details via Partition-then-Ensemble on Higher Resolution Images
	1 Introduction
	2 Proposed Method
		2.1 Two-Stage Architecture
		2.2 Training Algorithm
		2.3 Testing Algorithm
	3 Training Details
	4 Experiments and Results
	5 Conclusion
	References
Unconditionally Generated and Pseudo-Labeled Synthetic Images for Diabetic Foot Ulcer Segmentation Dataset Extension
	1 Introduction
	2 Dataset and Methods
		2.1 Diabetic Foot Ulcer Challenge 2022 Dataset
		2.2 Semantic Segmentation via Segmentation Models PyTorch
		2.3 Image Synthesis via StyleGAN2+ADA
		2.4 Experimental Environment
	3 Approach
		3.1 Pre-processing
		3.2 Augmentations
		3.3 Baseline Model Training and Ensemble
		3.4 Synthetic Image Generation and Dataset Extension
		3.5 Extended Model Training and Ensemble
		3.6 Inference and Post-processing
	4 Results
		4.1 Baseline and Extended Segmentation Ensemble Performance
		4.2 Synthetic Images and Pseudo-Labels for Dataset Extension
	5 Discussion
		5.1 Segmentation Performance and Post-processing
		5.2 Dataset Extension via Pseudo-Labeled Synthetic Images
		5.3 Limitations
	6 Conclusion
	References
Post Challenge Papers
Diabetic Foot Ulcer Segmentation Using Convolutional and Transformer-Based Models
	1 Introduction
	2 Background
	3 Related Work
	4 Dataset
	5 Proposed Method
		5.1 Models
		5.2 Training Procedure
		5.3 Testing Procedure
		5.4 Ensemble Models
	6 Results
	7 Conclusion
	References
Refined Mixup Augmentation for Diabetic Foot Ulcer Segmentation
	1 Introduction
	2 Dataset
	3 Methodology
		3.1 Preprocessing
		3.2 Proposed Approach
	4 Experimental Results
		4.1 Fault-Case Analysis
	5 Conclusions
	References
DFU-Ens: End-to-End Diabetic Foot Ulcer Segmentation Framework with Vision Transformer Based Detection
	1 Introduction
		1.1 Related Work
	2 DFU-Ens: End-to-End Ensemble DFU Segmentation Framework
		2.1 Dataset: Diabetic Foot Ulcer Segmentation Challenge 2022
		2.2 Module I: End-to-End Segmentation Framework Based on the U-Net Architecture
		2.3 Hybrid Solution Combining Detection and Patch Segmentation with YOLOv4 and Vision Transformers
	3 Experimental Results and Analysis
		3.1 Hardware and Software
	4 Conclusion
	References
Summary Paper
Diabetic Foot Ulcer Grand Challenge 2022 Summary
	1 Introduction
	2 Methodology
		2.1 Datasets and Ground Truth
		2.2 Performance Metrics
		2.3 Summary of the Proposed Methods
	3 Results and Discussion
	4 Conclusion
	References
Author Index