Clinical evaluation of super-resolution for brain MRI images based on generative adversarial networks
In magnetic resonance imaging (MRI), reducing long scan times is an urgent issue that could be addressed with super-resolution (SR) techniques. Most of the SR networks using deep neural networks (DNNs) have been evaluated only based on numeric metrics, and the image restoration quality for individua...
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2022-01-01
|
| Series: | Informatics in Medicine Unlocked |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2352914822001721 |
| _version_ | 1828419368751464448 |
|---|---|
| author | Yasuhiko Terada Tomoki Miyasaka Ai Nakao Satoshi Funayama Shintaro Ichikawa Tomohiro Takamura Daiki Tamada Hiroyuki Morisaka Hiroshi Onishi |
| author_facet | Yasuhiko Terada Tomoki Miyasaka Ai Nakao Satoshi Funayama Shintaro Ichikawa Tomohiro Takamura Daiki Tamada Hiroyuki Morisaka Hiroshi Onishi |
| author_sort | Yasuhiko Terada |
| collection | DOAJ |
| description | In magnetic resonance imaging (MRI), reducing long scan times is an urgent issue that could be addressed with super-resolution (SR) techniques. Most of the SR networks using deep neural networks (DNNs) have been evaluated only based on numeric metrics, and the image restoration quality for individual lesions is not evaluated sufficiently. Here, we evaluated the performances of different types of SR networks using DNNs for brain MRI, in terms of diagnostic performance and image quality. We focused on comparing the performance between generative adversarial networks (GANs) and non-GAN networks. There was a trade-off in such restoration quality between GAN- and non-GAN-based SRs, with the GANs being more accurate in restoring images of anatomical structures but less accurate in restoring those of lesions; non-GANs showed the opposite tendency. The non-GAN SRs were preferable in terms of diagnostic performance and image quality. This result suggested that the evaluation of DNN performance for lesions might be changed drastically by adding a clinical evaluation perspective. The dependence of network architecture on network performance obtained in this study will provide guidance for future development of SR DNN for medical images. |
| first_indexed | 2024-12-10T14:52:19Z |
| format | Article |
| id | doaj.art-428e9a21fb724e379a8da04166515ec0 |
| institution | Directory Open Access Journal |
| issn | 2352-9148 |
| language | English |
| last_indexed | 2024-12-10T14:52:19Z |
| publishDate | 2022-01-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Informatics in Medicine Unlocked |
| spelling | doaj.art-428e9a21fb724e379a8da04166515ec02022-12-22T01:44:25ZengElsevierInformatics in Medicine Unlocked2352-91482022-01-0132101030Clinical evaluation of super-resolution for brain MRI images based on generative adversarial networksYasuhiko Terada0Tomoki Miyasaka1Ai Nakao2Satoshi Funayama3Shintaro Ichikawa4Tomohiro Takamura5Daiki Tamada6Hiroyuki Morisaka7Hiroshi Onishi8Institute of Applied Physics, University of Tsukuba, Tsukuba, Japan; Corresponding author. Institute of Applied Physics, University of Tsukuba, Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki, 305-8573, Japan.Institute of Applied Physics, University of Tsukuba, Tsukuba, JapanInstitute of Applied Physics, University of Tsukuba, Tsukuba, JapanDepartment of Radiology, University of Yamanashi, Chuo, JapanDepartment of Radiology, Hamamatsu University School of Medicine, Hamamatsu, JapanDepartment of Radiology, Shizuoka General Hospital, Shizuoka, JapanDepartment of Radiology, University of Yamanashi, Chuo, JapanDepartment of Radiology, University of Yamanashi, Chuo, JapanDepartment of Radiology, University of Yamanashi, Chuo, JapanIn magnetic resonance imaging (MRI), reducing long scan times is an urgent issue that could be addressed with super-resolution (SR) techniques. Most of the SR networks using deep neural networks (DNNs) have been evaluated only based on numeric metrics, and the image restoration quality for individual lesions is not evaluated sufficiently. Here, we evaluated the performances of different types of SR networks using DNNs for brain MRI, in terms of diagnostic performance and image quality. We focused on comparing the performance between generative adversarial networks (GANs) and non-GAN networks. There was a trade-off in such restoration quality between GAN- and non-GAN-based SRs, with the GANs being more accurate in restoring images of anatomical structures but less accurate in restoring those of lesions; non-GANs showed the opposite tendency. The non-GAN SRs were preferable in terms of diagnostic performance and image quality. This result suggested that the evaluation of DNN performance for lesions might be changed drastically by adding a clinical evaluation perspective. The dependence of network architecture on network performance obtained in this study will provide guidance for future development of SR DNN for medical images.http://www.sciencedirect.com/science/article/pii/S2352914822001721Deep neural networksSuper-resolutionClinical evaluationGenerative adversarial networksBrain MRI |
| spellingShingle | Yasuhiko Terada Tomoki Miyasaka Ai Nakao Satoshi Funayama Shintaro Ichikawa Tomohiro Takamura Daiki Tamada Hiroyuki Morisaka Hiroshi Onishi Clinical evaluation of super-resolution for brain MRI images based on generative adversarial networks Informatics in Medicine Unlocked Deep neural networks Super-resolution Clinical evaluation Generative adversarial networks Brain MRI |
| title | Clinical evaluation of super-resolution for brain MRI images based on generative adversarial networks |
| title_full | Clinical evaluation of super-resolution for brain MRI images based on generative adversarial networks |
| title_fullStr | Clinical evaluation of super-resolution for brain MRI images based on generative adversarial networks |
| title_full_unstemmed | Clinical evaluation of super-resolution for brain MRI images based on generative adversarial networks |
| title_short | Clinical evaluation of super-resolution for brain MRI images based on generative adversarial networks |
| title_sort | clinical evaluation of super resolution for brain mri images based on generative adversarial networks |
| topic | Deep neural networks Super-resolution Clinical evaluation Generative adversarial networks Brain MRI |
| url | http://www.sciencedirect.com/science/article/pii/S2352914822001721 |
| work_keys_str_mv | AT yasuhikoterada clinicalevaluationofsuperresolutionforbrainmriimagesbasedongenerativeadversarialnetworks AT tomokimiyasaka clinicalevaluationofsuperresolutionforbrainmriimagesbasedongenerativeadversarialnetworks AT ainakao clinicalevaluationofsuperresolutionforbrainmriimagesbasedongenerativeadversarialnetworks AT satoshifunayama clinicalevaluationofsuperresolutionforbrainmriimagesbasedongenerativeadversarialnetworks AT shintaroichikawa clinicalevaluationofsuperresolutionforbrainmriimagesbasedongenerativeadversarialnetworks AT tomohirotakamura clinicalevaluationofsuperresolutionforbrainmriimagesbasedongenerativeadversarialnetworks AT daikitamada clinicalevaluationofsuperresolutionforbrainmriimagesbasedongenerativeadversarialnetworks AT hiroyukimorisaka clinicalevaluationofsuperresolutionforbrainmriimagesbasedongenerativeadversarialnetworks AT hiroshionishi clinicalevaluationofsuperresolutionforbrainmriimagesbasedongenerativeadversarialnetworks |