Current Result Document :
| 한글제목(Korean Title) |
의료 이미지 분할을 위한 일관성 기반의 선생-학생 준 지도 학습 프레임워크 |
| 영문제목(English Title) |
Consistency-based Teacher-Student Framework for Semi-supervised Medical Image Segmentation |
| 저자(Author) |
Nguyen P. Bui
김현성
Duc-Tai Le
추현승
Nguyen P. Bui
Hyunsung Kim
Duc-Tai Le
Hyunseung Choo
|
| 원문수록처(Citation) |
VOL 23 NO. 01 PP. 0199 ~ 0200 (2022. 04) |
한글내용
(Korean Abstract) |
|
영문내용
(English Abstract) |
Precise and robust segmentation for medical images is crucial for many clinical applications such as disease diagnosis and treatment planning. Medical image segmentation remains a challenging task due to many factors. With the development of deep learning, several methods have achieved great success in medical image segmentation. However, these methods require a large number of manual annotations, which are very scarce in the medical area. Semi-supervised learning can alleviate the need for labeled data by effectively exploiting the unlabeled data to train deep and complicated models. We present a strategy named consistency-based teacher-student (CoTS) framework for semi-supervised medical image segmentation. Our method consists of two networks: a pre-trained teacher network and a student network. CoTS improves the segmentation by minimizing the difference between the teacher and the student networks on both labeled and unlabeled data. We have extensively evaluated our proposed method on binary segmentation task for two 3D datasets, including left atrium-MRI and pancreas-CT. Compared with other semi-supervised methods, our CoTS has demonstrated superior performance.
|
| 키워드(Keyword) |
|
| 파일첨부 |
PDF 다운로드
|
