Abstract
It remains a mystery as to how neurons are connected and thereby enable use to think, and volume reconstruction from series of microscopy sections of brains is a vital technique in determining this connectivity. Image registration is a key component; the aim of image registration is to estimate the deformation field between two images. Current methods choose to directly regress the deformation field; however, this task is very challenging. It is common to trade off computational complexity with precision when designing complex models for deformation field estimation. This approach is very inefficient, leading to a long inference time. In this paper, we suggest that complex models are not necessary and solve this dilemma by proposing a dual-network architecture. We divide the deformation field prediction problem into two relatively simple subproblems and solve each of them on one branch of the proposed dual network. The two subproblems have completely opposite properties, and we fully utilize these properties to simplify the design of the dual network. These simple architectures enable high-speed image registration. The two branches are able to work together and make up for each other’s drawbacks, and no loss of accuracy occurs even when simple architectures are involved. Furthermore, we introduce a series of loss functions to enable the joint training of the two networks in an unsupervised manner without introducing costly manual annotations. The experimental results reveal that our method outperforms state-of-the-art methods in fly brain electron microscopy image registration tasks, and further ablation studies enable us to obtain a comprehensive understanding of each component of our network.
Highlights
Image registration is a long-standing problem in computer vision and medical image processing.The aim of image registration is to estimate the deformation field between two images and warp source images to reference images according to the estimated deformation field
The experimental results reveal that our method outperforms state-of-the-art methods in the fly brain electron microscopy image registration task, and further ablation studies enable us to obtain a comprehensive understanding of each component of our network
We aim to solve the image registration problem, which is a key component of volume reconstruction from brain microscopy sections
Summary
Image registration is a long-standing problem in computer vision and medical image processing. Since linear deformations pay greater attention to the global transformation of the image, they do not focus on local details. Utilizing this property, we can use low-resolution inputs to reduce the computational complexity. Since a large part of the deformation represented by the affine transformation has been eliminated, the remaining nonlinear deformation can be seen as small variations of fine details; such deformation focuses on local structures, no high-level semantic information is involved, and a shallow neural network is sufficient. The experimental results reveal that our method outperforms state-of-the-art methods in the fly brain electron microscopy image registration task, and further ablation studies enable us to obtain a comprehensive understanding of each component of our network
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.