Title Page
Contents
Chapter 1. Introduction 9
Chapter 2. Methods 13
2.1. Overall Framework 13
2.2. Data Preprocessing 16
2.3. Uncertainty-Aware Multi-Scale Prediction Network 17
2.4. Objective Functions 20
Chapter 3. Results 23
3.1. Datasets 23
3.2. Experimental Settings 24
3.3. Results 25
Chapter 4. Discussions 33
Chapter 5. Conclusions 35
References 36
국문요약 41
Table 1. Quantitative evaluation of pancreas segmentation results: The mean, standard deviation, minimum and maximum values of the... 26
Table 2. Quantitative evaluation between our method and other semi-supervised methods: The last row is the proposed method, and... 32
Figure 1. Characteristics of the pancreas on abdominal CT images: (a) Abdominal CT images and its annotation in axial planes: blurred... 10
Figure 2. Overview of the proposed semi-supervised segmentation of the pancreas through uncertainty-aware dual-task consistency: The... 14
Figure 3. Detail of the uncertainty-aware multi-scale prediction network (UA-MP Net): UA-MP Net is based on a backbone of 2D... 19
Figure 4. Qualitative evaluation of pancreas segmentation results: The red, blue, and green colors represent true positive, false positive, and... 29