Multi-dimensional consistency learning between 2D Swin U-Net and 3D U-Net for intestine segmentation from CT volume
The proposed method is based on semi-supervised learning and involves training both 2D Swin U-Net and 3D U-Net. The method mitigates the impact of limited labeled data and maintains consistncy of multi-dimensional outputs from the two networks to improve the segmentation accuracy. Compared to previous methods, our method demonstrates superior segmentation performance.
Authors
- Akinari Hinoki (ORCID: https://orcid.org/0000-0001-6878-2043)
- Takayuki Kitasaka (ORCID: https://orcid.org/0000-0003-1648-1999)
- Hiroo Uchida (ORCID: https://orcid.org/0000-0002-1818-6697)
- An Qin (ORCID: https://orcid.org/0000-0002-3333-5363)
- Kensaku Mori (ORCID: https://orcid.org/0000-0002-0100-4797)
- Yuichiro Hayashi (ORCID: https://orcid.org/0000-0001-5241-8669)
- Kojiro Suzuki (ORCID: https://orcid.org/0000-0002-8403-9425)
- Aitaro Takimoto (ORCID: https://orcid.org/0000-0002-8744-2949)
- Masahiro Oda (ORCID: https://orcid.org/0000-0001-7714-422X)
- Hirohisa Oda (ORCID: https://orcid.org/0000-0003-0896-4333)
Institutions
- University of Shizuoka (JP)
- Aichi Medical University (JP)
- National Institute of Informatics (JP)
- Aichi Institute of Technology (JP)
- Nagoya University (JP)
Publication Details
- Journal
- International Journal of Computer Assisted Radiology and Surgery
- Published
- 2025-02-22
- DOI
- https://doi.org/10.1007/s11548-024-03252-6
- Citations
- 1
- Primary Topic
- Colorectal Cancer Screening and Detection
- Type
- article
- Field-Weighted Citation Impact
- 1.05
Funders
- Ministry of Education, Culture, Sports, Science and Technology
- Hori Sciences and Arts Foundation
- Japan Society for the Promotion of Science
- Core Research for Evolutional Science and Technology