Dense motion propogation from sparse samples for free breathing respiratory motion modelling

Abstract

The development of models of respiration for motion correction in diagnostic imaging have often utilized dynamic 4D data, on which plausible respiratory motion patterns can be estimated or validated. To date, dynamic 4D MRI has often been used, its attraction lying in its zero radiation burden and large field of view. However, limitations in scanner technology produces poor contrast images when such volumetric data are acquired at high speed (< 1s). Therefore, in this latest work we provide the first demonstration that sparse sampling of dynamic MRI may be used as an alternative approach to successfully build high contrast, high resolution 4D models of free-breathing respiratory motion. This is achieved by constrained articulation of a high contrast/ high spatial resolution single static volume. The articulation is based on estimating, in the eigen domain, complete 4D motion vector fields from sparsely-sampled freebreathing dynamic MRI data, but constrained by an eigenbasis that characterizes a corresponding average 4D single respiratory cycle dataset. We demonstrate that this approach can provide equivalent motion vector fields compared to fully sampled 4D dynamic data, whilst preserving the corresponding high resolution/ high contrast inherent in the corresponding static image volume. Validation is performed on three 4D MRI volunteer datasets using 8 extracted slices from a fast 4D acquisition (0.7sec per volume). The estimated deformation fields from sparse sampling are compared to the fully sampled equivalents, resulting in an rms error of the order of 2mm, validating the approach. We also present exemplar 4D high contrast, high resolution articulated volunteer datasets using this methodology. This approach facilitates greater freedom in the acquisition of free breathing respiratory motion sequences which may be used to inform motion modeling methods in a range of imaging modalities.

Dense motion propagation from sparse samples

“…However, in this work we choose the simplest form, Principal Component Analysis (PCA). The high dimensional respiratory motion data that we observe can then be approximated by a linear manifold in eigen-space (Smith et al 2013a) and so provides a straightforward linear projection of the data onto the principal components which form the basis vectors of the new linear subspace.…”