Unsupervised Domain Adaptation for Low-Dose Computed Tomography Denoising
Deep neural networks have shown great improvements in low-dose computed tomography (CT) denoising. Early deep learning-based low-dose CT denoising algorithms were primarily based on supervised learning. However, supervised learning requires a large number of training samples, which is impractical in...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2022-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9969607/ |
| _version_ | 1828091678264655872 |
|---|---|
| author | Jaa-Yeon Lee Wonjin Kim Yebin Lee Ji-Yeon Lee Eunji Ko Jang-Hwan Choi |
| author_facet | Jaa-Yeon Lee Wonjin Kim Yebin Lee Ji-Yeon Lee Eunji Ko Jang-Hwan Choi |
| author_sort | Jaa-Yeon Lee |
| collection | DOAJ |
| description | Deep neural networks have shown great improvements in low-dose computed tomography (CT) denoising. Early deep learning-based low-dose CT denoising algorithms were primarily based on supervised learning. However, supervised learning requires a large number of training samples, which is impractical in real-world scenarios. To address this problem, we propose a novel unsupervised domain adaptation approach for low-dose CT denoising. This proposed framework adapts the network pretrained with paired low- and normal-dose phantom images (source domain) to denoise unlabeled low-dose human CT images (target domain). Our framework modifies the action of the domain classifier, enabling the denoising network to be adapted to the target domain. Furthermore, we introduce a new backpropagation method, which we call domain-independent weighted backpropagation. By combining these techniques, we demonstrate that the denoising network can be properly trained without using clinical clean CT images. The experimental results showed that our method exhibited better performance in terms of both objective and perceptual image qualities when compared with current unsupervised denoising algorithms. Our proposed domain adaptation represents a first-use case in the context of CT denoising problems, with the possibility of extension to other image restoration tasks. |
| first_indexed | 2024-04-11T06:17:55Z |
| format | Article |
| id | doaj.art-d4e5f49d39894ae599eb404558ff7e2b |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-11T06:17:55Z |
| publishDate | 2022-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-d4e5f49d39894ae599eb404558ff7e2b2022-12-22T04:41:01ZengIEEEIEEE Access2169-35362022-01-011012658012659210.1109/ACCESS.2022.32265109969607Unsupervised Domain Adaptation for Low-Dose Computed Tomography DenoisingJaa-Yeon Lee0https://orcid.org/0000-0003-3755-9521Wonjin Kim1Yebin Lee2Ji-Yeon Lee3Eunji Ko4Jang-Hwan Choi5https://orcid.org/0000-0001-9273-034XDivision of Mechanical and Biomedical Engineering, Graduate Program in System Health Science and Engineering, Ewha Womans University, Seodaemun-gu, Seoul, Republic of KoreaDivision of Mechanical and Biomedical Engineering, Graduate Program in System Health Science and Engineering, Ewha Womans University, Seodaemun-gu, Seoul, Republic of KoreaDivision of Mechanical and Biomedical Engineering, Graduate Program in System Health Science and Engineering, Ewha Womans University, Seodaemun-gu, Seoul, Republic of KoreaDivision of Mechanical and Biomedical Engineering, Graduate Program in System Health Science and Engineering, Ewha Womans University, Seodaemun-gu, Seoul, Republic of KoreaDivision of Mechanical and Biomedical Engineering, Graduate Program in System Health Science and Engineering, Ewha Womans University, Seodaemun-gu, Seoul, Republic of KoreaDivision of Mechanical and Biomedical Engineering, Graduate Program in System Health Science and Engineering, Ewha Womans University, Seodaemun-gu, Seoul, Republic of KoreaDeep neural networks have shown great improvements in low-dose computed tomography (CT) denoising. Early deep learning-based low-dose CT denoising algorithms were primarily based on supervised learning. However, supervised learning requires a large number of training samples, which is impractical in real-world scenarios. To address this problem, we propose a novel unsupervised domain adaptation approach for low-dose CT denoising. This proposed framework adapts the network pretrained with paired low- and normal-dose phantom images (source domain) to denoise unlabeled low-dose human CT images (target domain). Our framework modifies the action of the domain classifier, enabling the denoising network to be adapted to the target domain. Furthermore, we introduce a new backpropagation method, which we call domain-independent weighted backpropagation. By combining these techniques, we demonstrate that the denoising network can be properly trained without using clinical clean CT images. The experimental results showed that our method exhibited better performance in terms of both objective and perceptual image qualities when compared with current unsupervised denoising algorithms. Our proposed domain adaptation represents a first-use case in the context of CT denoising problems, with the possibility of extension to other image restoration tasks.https://ieeexplore.ieee.org/document/9969607/Low-dose computed tomography (LDCT) denoisinglow-dose CTdeep learningdomain adaptationunsupervised learning |
| spellingShingle | Jaa-Yeon Lee Wonjin Kim Yebin Lee Ji-Yeon Lee Eunji Ko Jang-Hwan Choi Unsupervised Domain Adaptation for Low-Dose Computed Tomography Denoising IEEE Access Low-dose computed tomography (LDCT) denoising low-dose CT deep learning domain adaptation unsupervised learning |
| title | Unsupervised Domain Adaptation for Low-Dose Computed Tomography Denoising |
| title_full | Unsupervised Domain Adaptation for Low-Dose Computed Tomography Denoising |
| title_fullStr | Unsupervised Domain Adaptation for Low-Dose Computed Tomography Denoising |
| title_full_unstemmed | Unsupervised Domain Adaptation for Low-Dose Computed Tomography Denoising |
| title_short | Unsupervised Domain Adaptation for Low-Dose Computed Tomography Denoising |
| title_sort | unsupervised domain adaptation for low dose computed tomography denoising |
| topic | Low-dose computed tomography (LDCT) denoising low-dose CT deep learning domain adaptation unsupervised learning |
| url | https://ieeexplore.ieee.org/document/9969607/ |
| work_keys_str_mv | AT jaayeonlee unsuperviseddomainadaptationforlowdosecomputedtomographydenoising AT wonjinkim unsuperviseddomainadaptationforlowdosecomputedtomographydenoising AT yebinlee unsuperviseddomainadaptationforlowdosecomputedtomographydenoising AT jiyeonlee unsuperviseddomainadaptationforlowdosecomputedtomographydenoising AT eunjiko unsuperviseddomainadaptationforlowdosecomputedtomographydenoising AT janghwanchoi unsuperviseddomainadaptationforlowdosecomputedtomographydenoising |