Deformer: Towards Displacement Field Learning for Unsupervised Medical Image Registration

Chen, Jiashun; Lu, Donghuan; Wang, Dong; Ning, Munan; Shi, Xinyu; Xu, Zhe; Zheng, Yefeng

doi:10.1007/978-3-031-16446-0_14

Cited by 11 publications

(9 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…VoxelMorph [1] and ULAE [16] models use the CNN to complete registration. DMR [10] model refines the deformation field into the form of multiple vectors.…”

Section: Experimental Results Of Brain Datasetsmentioning

confidence: 99%

“…The VM [1] model is commonly used as a baseline for deep learning registration methods. Unlike other methods that directly generate deformation fields based on deep learning, Eppenhof et al [20] and DMR [10] models refine the deformation field into multiple vectors. The DLIR [21] model performs experiments on datasets representing anatomical information of different organs to reflect the registration potential of the model.…”

Section: Experimental Results Of Lung Datasetsmentioning

confidence: 99%

“…Eppenhof et al [20] Pace et al [17] DLIR [21] DMR [10] VM [1] Elastix [19] Schmidt-Richberg et al [18] Compared with the DMR model, our method reduces the average TRE from 2.40 mm to 1.99 mm under the condition that the negative percentage of Jacobian determinant and registration time is equivalent. It shows that the proposed model and deformation strategy can effectively represent the mapping of the image to spatial transformation and complete the prediction of the deformation field in a coarse-to-fine manner.…”

Section: Initialmentioning

confidence: 99%

“…Chen et al [10] proposed that the displacement in registration can be seen as a weighted sum of a set of fundamental vectors. The weight of each fundamental basis can be learned through cascaded feature maps.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Weakly supervised medical image registration with multi-information guidance

Liu,

Ren,

2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

In recent years, the registration method based on deep learning has received extensive attention from scholars due to its superiority in real-time performance. Most of the work directly use convolutional neural networks to map the image to be registered into the transform space. However, the receptive field of convolutional neural networks is limited, and multi-layer convolution superposition is needed to obtain a relatively large receptive field. Transformer-based methods can better express spatial relationships through attention mechanisms. However, the self-attention and the multi-head mechanisms make each small block calculate the relationship with other small blocks regardless of distance. Due to the limited moving range of corresponding voxel points in the medical images, this long-distance dependence may cause the model to be interfered by long-distance voxels. In this paper, we convert the spatial transformation of the corresponding voxel points into the calculation of the basic vector basis to propose the SV-basis module and design a two-stage multi-scale registration model. Experiments are carried out on brain and lung datasets to prove the effectiveness and universality of the proposed registration method. According to the anatomical characteristics of medical images, the corresponding loss function is designed to introduce mask information into the registration task. The experimental results show that the proposed method can accurately register brain and lung images.

show abstract

“…VoxelMorph [1] and ULAE [16] models use the CNN to complete registration. DMR [10] model refines the deformation field into the form of multiple vectors.…”

Section: Experimental Results Of Brain Datasetsmentioning

confidence: 99%

Section: Experimental Results Of Lung Datasetsmentioning

confidence: 99%

Section: Initialmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Weakly supervised medical image registration with multi-information guidance

Liu,

Ren,

2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…Considering that the core of registration is the alignment of anatomical structures between images, distant correlations in one image may not necessarily exist in another image, especially in cases of large-scale deformations. To address this, Che et al [5] proposed a novel Deformer module that predicts deformation fields by constructing a weighted sum of deformation bases, achieving robust mapping from image feature representation to spatial transformation. Zhu et al [6] further introduced the symmetrical Swin Transformer into unsupervised registration networks.…”

Section: Registration Methods For Large-scale Deformationmentioning

confidence: 99%

Spatial structure inter-correlation-based brain images registration

He,

Guan,

Huang

et al. 2024

International Conference on Computer Graphics, Artificial Intelligence, and Data Processing (ICCAID 2023)

View full text Add to dashboard Cite

Brain medical image registration is the foundation of numerous clinical applications in neurosurgery, such as preoperative planning and intraoperative navigation. However, precise registration remains challenging when dealing with large-scale deformations between images to be registered and significant anatomical differences across modalities. Existing methods often incorporate explicit structure information (e.g., anatomical structure segmentation masks) as prior knowledge to assist registration, but obtaining high-precision structure prior knowledge itself is difficult. In addition, most existing methods for large-scale registration use low-resolution advanced semantic relationships as initial driving information for registration, which may mislead the registration process that relies primarily on structure alignment. To address these issues, this paper proposes a novel brain image registration method based on spatial structure inter-correlation. Specifically, we first extract shallow multi-scale anatomical structure information because the core of registration is the alignment of corresponding structures. Compared to deep structures rich in semantic information, shallow structures rich in structure information are more conducive to guiding the registration process. Then, we construct structure space inter-correlation based on the extracted shallow multi-scale anatomical structure information (including structure attention calculation between two modalities). Different scales of structure information contain different levels of structure inter-correlation, and by performing multi-level low-resolution upsampling and fusion, we obtain the final registration result to improve registration accuracy. Finally, starting from dual registration, the deformation from moving images to fixed images should enable fixed images to deform to moving images in the dual direction, thereby improving the robustness of registration. Through comparison with many advanced methods on brain MRI datasets, our proposed method demonstrates superior accuracy in brain images registration.

show abstract