Deformable registration of chest CT images using a 3D convolutional neural network based on unsupervised learning

Abstract

The deformable registration of 3D chest computed tomography (CT) images is one of the most important tasks in the field of medical image registration. However, the nonlinear deformation and large-scale displacement of lung tissues caused by respiratory motion cause great challenges in the deformable registration of 3D lung CT images. Materials and methods: We proposed an end-to-end fast registration method based on unsupervised learning, optimized the classic U-Net, and added inception modules between skip connections. The inception module attempts to capture and merge information at different spatial scales to generate a highprecision dense displacement vector field. To solve the problem of voxel folding in flexible registration, we put the Jacobian regularization term into the loss function to directly penalize the singularity of the displacement field during training to ensure a smooth displacement vector field. In the stage of data preprocessing, we segmented the lung fields to eliminate the interference of irrelevant information in the network during training. The existing publicly available datasets cannot implement model training. To alleviate the problem of overfitting caused by limited data resources being available, we proposed a data augmentation method based on the 3D-TPS (3D thin plate spline) transform to expand the training data. Results: Compared with the experimental results obtained by using the Vox-elMorph deep learning method and registration packages, such as ANTs and Elastix, we achieved a competitive target registration error of 2.09 mm, an optimal Dice score of 0.987, and almost no folding voxels. Additionally, the proposed method was much faster than the traditional methods. Conclusions: In this study, we have shown that the proposed method was efficient in 3D chest image registration. The promising results demonstrated that our method showed strong robustness in the deformable registration of 3D chest CT images.

Deformable 3D medical image registration with convolutional neural network and transformer

“…They conclude that a more enormous weight will effectively reduce the number of foldings, but it will sacrifice a little registration accuracy. Zheng et al offered a fast registration network that optimized the classic U-Net in [18]. The negative Jacobian determinant regularization term into the loss function is also used to penalize the foldings of the displacement field directly.…”

“…The negative Jacobian determinant regularization term into the loss function is also used to penalize the foldings of the displacement field directly. For the research in [17,18], their loss functions can be expressed as…”

4D‐CT deformable image registration using unsupervised recursive cascaded full‐resolution residual networks

“…The DL-based DIR methods have been reported to have superior performance in brain MR images, head/neck CT images, chest CT images, lung 4D-CT images, and more. [14][15][16][17] According to the output of the network, the deep learning-based DIR methods could be divided into two categories: (1) DL-based similarity calculation; and (2) DLbased DVF prediction. The DL-based similarity calculation method was normally developed for multi-modality image registration.…”