A Lightweight Ultrasound Image Denoiser Using Parallel Attention Modules and Capsule Generative Adversarial Network
The quality of ultrasound (US) imaging has been constrained by its limited contrast and resolution, inherent speckle noise, and the presence of other artifacts. Existing traditional and deep learning-based US denoising approaches have many limitations, such as reliance on manual parameter configurat...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2024-01-01
|
| Series: | Informatics in Medicine Unlocked |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2352914824001254 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850210022432178176 |
|---|---|
| author | Anparasy Sivaanpu Kumaradevan Punithakumar Kokul Thanikasalam Michelle Noga Rui Zheng Dean Ta Edmond H.M. Lou Lawrence H. Le |
| author_facet | Anparasy Sivaanpu Kumaradevan Punithakumar Kokul Thanikasalam Michelle Noga Rui Zheng Dean Ta Edmond H.M. Lou Lawrence H. Le |
| author_sort | Anparasy Sivaanpu |
| collection | DOAJ |
| description | The quality of ultrasound (US) imaging has been constrained by its limited contrast and resolution, inherent speckle noise, and the presence of other artifacts. Existing traditional and deep learning-based US denoising approaches have many limitations, such as reliance on manual parameter configurations, poor performance for unknown noise levels, the requirement for a large number of training data, and high computational expense. To address these challenges, we propose a novel Generative Adversarial Network (GAN) based denoiser. Capsule networks are utilized in both the generator and discriminator of the proposed GAN to capture intricate sparse features with less complexity. In addition, bias components are removed from all neurons of the generator to handle the unknown noise levels. A parallel attention module is also included in the proposed model to further enhance denoising performance. The proposed approach is trained in a semi-supervised manner and can thus be trained with fewer labeled images. Experimental evaluation on publicly available HC18 and BUSI datasets showed that the proposed approach achieved state-of-the-art denoising performance, with PSNR values of 33.86 and 34.16, and SSIM indices of 0.91 and 0.90, respectively. Moreover, experiments showed that the proposed approach is lightweight and more than twice as fast as similar denoisers. |
| format | Article |
| id | doaj-art-379920f6833e49bb8956784b7bf8fda1 |
| institution | OA Journals |
| issn | 2352-9148 |
| language | English |
| publishDate | 2024-01-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Informatics in Medicine Unlocked |
| spelling | doaj-art-379920f6833e49bb8956784b7bf8fda12025-08-20T02:09:52ZengElsevierInformatics in Medicine Unlocked2352-91482024-01-015010156910.1016/j.imu.2024.101569A Lightweight Ultrasound Image Denoiser Using Parallel Attention Modules and Capsule Generative Adversarial NetworkAnparasy Sivaanpu0Kumaradevan Punithakumar1Kokul Thanikasalam2Michelle Noga3Rui Zheng4Dean Ta5Edmond H.M. Lou6Lawrence H. Le7Department of Radiology and Diagnostic Imaging, University of Alberta, Edmonton, Alberta, CanadaDepartment of Radiology and Diagnostic Imaging, University of Alberta, Edmonton, Alberta, CanadaDepartment of Computer Science, University of Jaffna, Jaffna, Sri LankaDepartment of Radiology and Diagnostic Imaging, University of Alberta, Edmonton, Alberta, CanadaSchool of Information Science and Technology, ShanghaiTech University, Shanghai, 201210, ChinaDepartment of Biomedical Engineering, School of Information Science and Technology, Fudan University, Shanghai, China; State Key Laboratory of Integrated Chips and Systems, Fudan University, Shanghai, ChinaDepartment of Electrical and Computer Engineering, University of Alberta, Edmonton, Alberta, CanadaDepartment of Radiology and Diagnostic Imaging, University of Alberta, Edmonton, Alberta, Canada; State Key Laboratory of Integrated Chips and Systems, Fudan University, Shanghai, China; Corresponding author at: Department of Radiology and Diagnostic Imaging, University of Alberta, Edmonton, Alberta, Canada.The quality of ultrasound (US) imaging has been constrained by its limited contrast and resolution, inherent speckle noise, and the presence of other artifacts. Existing traditional and deep learning-based US denoising approaches have many limitations, such as reliance on manual parameter configurations, poor performance for unknown noise levels, the requirement for a large number of training data, and high computational expense. To address these challenges, we propose a novel Generative Adversarial Network (GAN) based denoiser. Capsule networks are utilized in both the generator and discriminator of the proposed GAN to capture intricate sparse features with less complexity. In addition, bias components are removed from all neurons of the generator to handle the unknown noise levels. A parallel attention module is also included in the proposed model to further enhance denoising performance. The proposed approach is trained in a semi-supervised manner and can thus be trained with fewer labeled images. Experimental evaluation on publicly available HC18 and BUSI datasets showed that the proposed approach achieved state-of-the-art denoising performance, with PSNR values of 33.86 and 34.16, and SSIM indices of 0.91 and 0.90, respectively. Moreover, experiments showed that the proposed approach is lightweight and more than twice as fast as similar denoisers.http://www.sciencedirect.com/science/article/pii/S2352914824001254Capsule NetworkGenerative Adversarial NetworkImage denoisingSemi-supervisionUltrasound image |
| spellingShingle | Anparasy Sivaanpu Kumaradevan Punithakumar Kokul Thanikasalam Michelle Noga Rui Zheng Dean Ta Edmond H.M. Lou Lawrence H. Le A Lightweight Ultrasound Image Denoiser Using Parallel Attention Modules and Capsule Generative Adversarial Network Informatics in Medicine Unlocked Capsule Network Generative Adversarial Network Image denoising Semi-supervision Ultrasound image |
| title | A Lightweight Ultrasound Image Denoiser Using Parallel Attention Modules and Capsule Generative Adversarial Network |
| title_full | A Lightweight Ultrasound Image Denoiser Using Parallel Attention Modules and Capsule Generative Adversarial Network |
| title_fullStr | A Lightweight Ultrasound Image Denoiser Using Parallel Attention Modules and Capsule Generative Adversarial Network |
| title_full_unstemmed | A Lightweight Ultrasound Image Denoiser Using Parallel Attention Modules and Capsule Generative Adversarial Network |
| title_short | A Lightweight Ultrasound Image Denoiser Using Parallel Attention Modules and Capsule Generative Adversarial Network |
| title_sort | lightweight ultrasound image denoiser using parallel attention modules and capsule generative adversarial network |
| topic | Capsule Network Generative Adversarial Network Image denoising Semi-supervision Ultrasound image |
| url | http://www.sciencedirect.com/science/article/pii/S2352914824001254 |
| work_keys_str_mv | AT anparasysivaanpu alightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT kumaradevanpunithakumar alightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT kokulthanikasalam alightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT michellenoga alightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT ruizheng alightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT deanta alightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT edmondhmlou alightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT lawrencehle alightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT anparasysivaanpu lightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT kumaradevanpunithakumar lightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT kokulthanikasalam lightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT michellenoga lightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT ruizheng lightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT deanta lightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT edmondhmlou lightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork AT lawrencehle lightweightultrasoundimagedenoiserusingparallelattentionmodulesandcapsulegenerativeadversarialnetwork |