Deformable Slice-to-Volume Registration for Motion Correction of Fetal Body and Placenta MRI

Abstract

In MRI, motion correction for fetal body poses a particular challenge due to the presence of local non-rigid transformations of organs caused by bending and stretching. The existing slice-to-volume (SVR) reconstruction methods provide efficient solution for the fetal brain that undergoes only rigid transformation or 4D fetal heart with rigid states correlated to cardiac phases. However, for fetal body reconstruction, rigid registration cannot resolve the issue of misregistrations due to deformable motion. This results in propagation of registration error to the reconstructed volume and subsequent degradation of features. We propose a novel approach for non-rigid motion correction in 3D volumes based on an extension of the classical SVR method with hierarchical deformable registration scheme and structure-based outlier rejection. Deformable SVR (DSVR) method allows high resolution reconstruction of the fetal trunk and the robust scheme for structure-based rejection of misregistered slices minimises the impact of registration error. The method performance is evaluated by comparison to the SVR and patch-to-volume registration methods for reconstruction of fetal trunk on a series of fetal MRI datasets from 28-30 weeks gestational age (GA) range with varying degree of motion corruption. An additional phantom study with simulated nonrigid motion is used for the assessment of consistency of DSVR reconstructed volumes.

Antenatal thymus volumes in fetuses that delivered <32 weeks' gestation: An MRI pilot study

“…were obtained from 4-8 motion corrupted T2 MRI stacks using the deformable slice-to-volume reconstruction method. [13][14][15] Scan length did not exceed 1 hour.…”

“…The following scanning parameters were used: TR = 25 991 ms, TE = 80 ms, slice thickness of 2.5 mm, slice overlap of 1.25 × 1.25 × 1.25 mm on 1.5T, and 1.21 × 1.21 × 1.5 or 1.25 × 1.25 × 2.5 mm on 3T, with a flip angle = 90°. Three-dimensional (3D) MR images of the fetal thoraxwere obtained from 4-8 motion corrupted T2 MRI stacks using the deformable slice-to-volume reconstruction method [13][14][15]. Scan length did not exceed 1 hour.…”

Emerging magnetic resonance imaging techniques in open spina bifida in utero

“…For reconstruction of the foetal spine, internal organs and placenta, the SVR methods mentioned above cannot resolve the issue of difficult registrations due to non-rigid motion, resulting in degradation of image quality within the reconstructed volume. Therefore, a method was developed that uses non-rigid image registration and SVR (deformable SVR) which shows great promise for reconstruction high resolution images of the spine and surrounding organs in spina bifida [ 51 ]. Patch-to-volume reconstruction is also able to reconstruct a large field of view of non-rigidly deforming structures, using a parallelised patchwork optimisation (patch to volume reconstruction is used instead of SVR), SR, and automatic outlier rejection.…”

“…In spina bifida imaging, foetal MRI is greatly limited by the lower spatial resolution for the assessment of the level of the lesion and to depict the myelon. Recent works demonstrated that isotropic reconstructions of the entire foetal body including the spine are possible [ 51 ], which could be used to tackle this problem. Faster image acquisition would also improve the clinical utility of foetal MRI, for example, by using compressed sensing technique [ 54 ].…”

APPLAUSE: Automatic Prediction of PLAcental health via U-net Segmentation and statistical Evaluation

“…However, this could be expanded with further post-processing steps. These include placental 3D reconstruction, eg from orthogonal slice stacks or multiple dynamics as has been recently proposed for T2* data for the brain (41) and the placenta (28) or inter-slice motion correction techniques (42) to improve the 3D continuity of the data thus enabling 3D patch-based segmentation and/or placental flattening techniques (16, 43), resulting in representations in a common coordinate system.…”