MSCAReg‐Net: Multi‐scale complexity‐aware convolutional neural network for deformable image registration
Abstract Deep learning‐based image registration (DLIR) has been widely developed, but it remains challenging in perceiving small and large deformations. Besides, the effectiveness of the DLIR methods was also rarely validated on the downstream tasks. In the study, a multi‐scale complexity‐aware regi...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2024-03-01
|
| Series: | IET Image Processing |
| Subjects: | |
| Online Access: | https://doi.org/10.1049/ipr2.12988 |
| _version_ | 1797271924075659264 |
|---|---|
| author | Hu Yu Qiang Zheng Fang Hu Chaoqing Ma Shuo Wang Shuai Wang |
| author_facet | Hu Yu Qiang Zheng Fang Hu Chaoqing Ma Shuo Wang Shuai Wang |
| author_sort | Hu Yu |
| collection | DOAJ |
| description | Abstract Deep learning‐based image registration (DLIR) has been widely developed, but it remains challenging in perceiving small and large deformations. Besides, the effectiveness of the DLIR methods was also rarely validated on the downstream tasks. In the study, a multi‐scale complexity‐aware registration network (MSCAReg‐Net) was proposed by devising a complexity‐aware technique to facilitate DLIR under a single‐resolution framework. Specifically, the complexity‐aware technique devised a multi‐scale complexity‐aware module (MSCA‐Module) to perceive deformations with distinct complexities, and employed a feature calibration module (FC‐Module) and a feature aggregation module (FA‐Module) to facilitate the MSCA‐Module by generating more distinguishable deformation features. Experimental results demonstrated the superiority of the proposed MSCAReg‐Net over the existing methods in terms of registration accuracy. Besides, other than the indices of Dice similarity coefficient (DSC) and percentage of voxels with non‐positive Jacobian determinant (|Jϕ|≤0), a comprehensive evaluation of the registration performance was performed by applying this method on a downstream task of multi‐atlas hippocampus segmentation (MAHS). Experimental results demonstrated that this method contributed to a better hippocampus segmentation over other DLIR methods, and a comparable segmentation performance with the leading SyN method. The comprehensive assessment including DSC, |Jϕ|≤0, and the downstream application on MAHS demonstrated the advances of this method. |
| first_indexed | 2024-03-07T14:17:09Z |
| format | Article |
| id | doaj.art-1b2e06564c7245fcbfc7d52534e5cc62 |
| institution | Directory Open Access Journal |
| issn | 1751-9659 1751-9667 |
| language | English |
| last_indexed | 2024-03-07T14:17:09Z |
| publishDate | 2024-03-01 |
| publisher | Wiley |
| record_format | Article |
| series | IET Image Processing |
| spelling | doaj.art-1b2e06564c7245fcbfc7d52534e5cc622024-03-06T11:42:57ZengWileyIET Image Processing1751-96591751-96672024-03-0118483985510.1049/ipr2.12988MSCAReg‐Net: Multi‐scale complexity‐aware convolutional neural network for deformable image registrationHu Yu0Qiang Zheng1Fang Hu2Chaoqing Ma3Shuo Wang4Shuai Wang5School of Computer and Control Engineering Yantai University Yantai Shandong Province ChinaSchool of Computer and Control Engineering Yantai University Yantai Shandong Province ChinaKey Laboratory of Medical Imaging and Artificial Intelligence of Hunan Province Xiangnan University Chenzhou Hunan Province ChinaSchool of Computer and Control Engineering Yantai University Yantai Shandong Province ChinaYantai University Trier College of Sustainable Technology Yantai University Yantai Shandong Province ChinaDepartment of Radiology Binzhou Medical University Hospital Binzhou Shandong Province ChinaAbstract Deep learning‐based image registration (DLIR) has been widely developed, but it remains challenging in perceiving small and large deformations. Besides, the effectiveness of the DLIR methods was also rarely validated on the downstream tasks. In the study, a multi‐scale complexity‐aware registration network (MSCAReg‐Net) was proposed by devising a complexity‐aware technique to facilitate DLIR under a single‐resolution framework. Specifically, the complexity‐aware technique devised a multi‐scale complexity‐aware module (MSCA‐Module) to perceive deformations with distinct complexities, and employed a feature calibration module (FC‐Module) and a feature aggregation module (FA‐Module) to facilitate the MSCA‐Module by generating more distinguishable deformation features. Experimental results demonstrated the superiority of the proposed MSCAReg‐Net over the existing methods in terms of registration accuracy. Besides, other than the indices of Dice similarity coefficient (DSC) and percentage of voxels with non‐positive Jacobian determinant (|Jϕ|≤0), a comprehensive evaluation of the registration performance was performed by applying this method on a downstream task of multi‐atlas hippocampus segmentation (MAHS). Experimental results demonstrated that this method contributed to a better hippocampus segmentation over other DLIR methods, and a comparable segmentation performance with the leading SyN method. The comprehensive assessment including DSC, |Jϕ|≤0, and the downstream application on MAHS demonstrated the advances of this method.https://doi.org/10.1049/ipr2.12988biomedical MRIcomputer visionconvolutional neural netsimage registrationmedical image processingunsupervised learning |
| spellingShingle | Hu Yu Qiang Zheng Fang Hu Chaoqing Ma Shuo Wang Shuai Wang MSCAReg‐Net: Multi‐scale complexity‐aware convolutional neural network for deformable image registration IET Image Processing biomedical MRI computer vision convolutional neural nets image registration medical image processing unsupervised learning |
| title | MSCAReg‐Net: Multi‐scale complexity‐aware convolutional neural network for deformable image registration |
| title_full | MSCAReg‐Net: Multi‐scale complexity‐aware convolutional neural network for deformable image registration |
| title_fullStr | MSCAReg‐Net: Multi‐scale complexity‐aware convolutional neural network for deformable image registration |
| title_full_unstemmed | MSCAReg‐Net: Multi‐scale complexity‐aware convolutional neural network for deformable image registration |
| title_short | MSCAReg‐Net: Multi‐scale complexity‐aware convolutional neural network for deformable image registration |
| title_sort | mscareg net multi scale complexity aware convolutional neural network for deformable image registration |
| topic | biomedical MRI computer vision convolutional neural nets image registration medical image processing unsupervised learning |
| url | https://doi.org/10.1049/ipr2.12988 |
| work_keys_str_mv | AT huyu mscaregnetmultiscalecomplexityawareconvolutionalneuralnetworkfordeformableimageregistration AT qiangzheng mscaregnetmultiscalecomplexityawareconvolutionalneuralnetworkfordeformableimageregistration AT fanghu mscaregnetmultiscalecomplexityawareconvolutionalneuralnetworkfordeformableimageregistration AT chaoqingma mscaregnetmultiscalecomplexityawareconvolutionalneuralnetworkfordeformableimageregistration AT shuowang mscaregnetmultiscalecomplexityawareconvolutionalneuralnetworkfordeformableimageregistration AT shuaiwang mscaregnetmultiscalecomplexityawareconvolutionalneuralnetworkfordeformableimageregistration |