Regular Article: Computing Fourier Transforms and Convolutions on the 2-Sphere
Advances in Applied Mathematics
| Hi-index | 0.00 |
Spherical navigators are an attractive approach to motion compensation in Magnetic Resonance Imaging. Because they can be acquired quickly, spherical navigators have the potential to measure and correct for rigid motion during image acquisition (prospectively as opposed to retrospectively). A limiting factor to prospective use of navigators is the time required to estimate the motion parameters. This estimation problem can be separated into a rotational and translational component. Recovery of the rotational motion can be cast as a registration of functions defined on a sphere. Previous methods for solving this registration problem are based on optimization strategies that are iterative and require k-space interpolation. Such approaches have undesirable convergence behavior for prospective use since the estimation complexity depends on both the number of samples and the amount of rotation. We propose and demonstrate an efficient algorithm for recovery of rotational motion using spherical navigators. We decompose the navigator magnitude using the spherical harmonic transform. In this framework, rigid rotations can be recovered from an over-constrained system of equations, leading to a computationally efficient algorithm for prospective motion compensation. The resulting algorithm is compared to existing approaches in simulated and actual navigator data. These results show that the spherical harmonic based estimation algorithm is significantly faster than existing methods and so is suited for prospective motion correction.