Accelerating multi-scale flows for LDDKBM diffeomorphic registration

TY - GEN

T1 - Accelerating multi-scale flows for LDDKBM diffeomorphic registration

AU - Sommer, Stefan Horst

PY - 2011

Y1 - 2011

N2 - Registrations in medical imaging and computational anatomy can be obtained using the Large Deformation Diffeomorphic Kernel Bundle Mapping (LDDKBM) framework. This provides a registration algorithm with a solid mathematical foundation while incorporating regularization of deformation at multiple scales. Because the variational formulation of LDDKBM implies a heavy computational burden in the search for optimal registrations, exploiting every possibility for faster computation will improve the usability of the algorithm. We present a parallelization strategy using the multi-scale structure and show that the parallelized method constitutes an example of how the processing power of GPUs can massively reduce the running time: after moving the computation to the GPU, we achieve a two order of magnitude speedup over a single-threaded CPU implementation. Not only does this significantly reduce the cost of using multiple scales, it also allows the algorithm to be used on much larger datasets.

AB - Registrations in medical imaging and computational anatomy can be obtained using the Large Deformation Diffeomorphic Kernel Bundle Mapping (LDDKBM) framework. This provides a registration algorithm with a solid mathematical foundation while incorporating regularization of deformation at multiple scales. Because the variational formulation of LDDKBM implies a heavy computational burden in the search for optimal registrations, exploiting every possibility for faster computation will improve the usability of the algorithm. We present a parallelization strategy using the multi-scale structure and show that the parallelized method constitutes an example of how the processing power of GPUs can massively reduce the running time: after moving the computation to the GPU, we achieve a two order of magnitude speedup over a single-threaded CPU implementation. Not only does this significantly reduce the cost of using multiple scales, it also allows the algorithm to be used on much larger datasets.

U2 - 10.1109/ICCVW.2011.6130284

DO - 10.1109/ICCVW.2011.6130284

M3 - Article in proceedings

SN - 978-1-4673-0062-9

SP - 499

EP - 505

BT - 2011 IEEE International Conference on Computer Vision Workshops (ICCV Workshops)

PB - IEEE

T2 - ICCV2011 Workshop

Y2 - 11 November 2011 through 11 November 2011

ER -