Purpose: Imaging registration has a significant contribution to guide and support physicians in the process of decision-making for diagnosis, prognosis, and treatment. However, existing registration methods based on the convolutional neural network cannot extract global features effectively, which significantly influences registration performance. Moreover, the smoothness of the displacement vector field (DVF) fails to be ensured due to the miss folding penalty.
Methods: In order to capture abundant global information as well as local information, we have proposed a novel 3D deformable image registration network based on Transformer (TransDIR). In the encoding phase, the transformer with the atrous reduction attention block is designed to capture the long-distance dependencies that are crucial for extracting global information. A zero-padding position encoder is embedded into the transformer to capture the local information. In the decoding phase, an up-sampling module based on an attention mechanism is designed to increase the significance of ROIs. Because of adding folding penalty term into loss function, the smoothness of DVF is improved.
Results: Finally, we carried out experiments on OASIS, LPBA40, MGH10, and MM-WHS open datasets to validate the effectiveness of TransDIR. Compared with LapIRN, the DSC score is improved by 1.1% and 0.9% on OASIS and LPBA40, separately. In addition, compared with VoxelMorph, the DSC score is improved by 2.8% on the basis of the folding index decreased by hundreds of times on MM-WHS.
Conclusions: The results show that the TransDIR achieves robust registration and promising generalizability compared with LapIRN and VoxelMorph.
Keywords: global feature extraction; registration; transformer; zero padding.
© 2021 American Association of Physicists in Medicine.