A matrix-free approach to parallel and memory-efficient deformable image registration
From MaRDI portal
(Redirected from Publication:4568099)
Abstract: We present a novel computational approach to fast and memory-efficient deformable image registration. In the variational registration model, the computation of the objective function derivatives is the computationally most expensive operation, both in terms of runtime and memory requirements. In order to target this bottleneck, we analyze the matrix structure of gradient and Hessian computations for the case of the normalized gradient fields distance measure and curvature regularization. Based on this analysis, we derive equivalent matrix-free closed-form expressions for derivative computations, eliminating the need for storing intermediate results and the costs of sparse matrix arithmetic. This has further benefits: (1) matrix computations can be fully parallelized, (2) memory complexity for derivative computation is reduced from linear to constant, and (3) overall computation times are substantially reduced. In comparison with an optimized matrix-based reference implementation, the CPU implementation achieves speedup factors between 3.1 and 9.7, and we are able to handle substantially higher resolutions. Using a GPU implementation, we achieve an additional speedup factor of up to 9.2. Furthermore, we evaluated the approach on real-world medical datasets. On ten publicly available lung CT images from the DIR-Lab 4DCT dataset, we achieve the best mean landmark error of 0.93 mm compared to other submissions on the DIR-Lab website, with an average runtime of only 9.23 s. Complete non-rigid registration of full-size 3D thorax-abdomen CT volumes from oncological follow-up is achieved in 12.6 s. The experimental results show that the proposed matrix-free algorithm enables the use of variational registration models also in applications which were previously impractical due to memory or runtime restrictions.
Recommendations
- scientific article; zbMATH DE number 2152736
- Efficient variational approaches for deformable registration of images
- scientific article; zbMATH DE number 1943007
- Variational image registration using inhomogeneous regularization
- CLAIRE: a distributed-memory solver for constrained large deformation diffeomorphic image registration
Cites work
- A fast full multigrid solver for applications in image processing
- A hyperelastic regularization energy for image registration
- A semi-Lagrangian two-level preconditioned Newton-Krylov solver for constrained diffeomorphic image registration
- Adaptive Mesh Refinement for Nonparametric Image Registration
- An Octree Method for Parametric Image Registration
- Curvature based image registration
- FAIR. Flexible algorithms for image registration.
- scientific article; zbMATH DE number 1943007 (Why is no real title available?)
- scientific article; zbMATH DE number 2021122 (Why is no real title available?)
- Image registration for remote sensing
- Numerical Optimization
Cited in
(4)- CLAIRE: a distributed-memory solver for constrained large deformation diffeomorphic image registration
- A structured L-BFGS method and its application to inverse problems
- A structured L-BFGS method with diagonal scaling and its application to image registration
- Hessian initialization strategies for \(\ell \)-BFGS solving non-linear inverse problems
This page was built for publication: A matrix-free approach to parallel and memory-efficient deformable image registration
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4568099)
