Planar-motion correction with use of K-space data acquired in Fourier MR imaging

The authors have developed a method for reducing magnetic resonance (MR) image artifacts caused by planar motion. Segments of k-space acquired with the subject stationary are detected automatically. Each k-space segment is Fourier transformed into an image in which rotational and translational displacements are measured manually. Before correction, k-space is made as Hermitian as allowed by the largest symmetric range of low spatial frequencies acquired with the subject stationary. Segments of k-space acquired with the subject in different positions are corrected separately. Although translation corrections can be applied effectively to both k-space and the spatial domain, rotation corrections are applied in the spatial domain to avoid image artifacts. To complement the correction, data corrupted by rotation are replaced by the complex conjugate of data of the opposite kappa x and kappa y, provided that these data have not been corrupted by rotation. The method reduced ghosts and blurring substantially on sagittal head images acquired with a standard spin-echo pulse sequence while a volunteer subject nodded his head.