View angle tilting echo planar imaging for distortion correction

Abstract

Geometric distortion caused by field inhomogeneity along the phase-encode direction is one of the most prominent artifacts due to a relatively low effective bandwidth along that direction in magnetic resonance echo planar imaging. This work describes a method for correcting in-plane image distortion along the phase-encode direction using a view angle tilting imaging technique in spin-echo echo planar imaging. Spin-echo echo planar imaging with view angle tilting uses the addition of gradient blips along the slice-select direction, concurrently applied with the phase-encode gradient blips, producing an additional phase. This phase effectively offsets an unwanted phase accumulation caused by field inhomogeneity, resulting in the removal of image distortion along the phase-encode direction. The proposed method is simple and straightforward both in implementation and application with no scan time penalty. Therefore, it is readily applicable on commercial scanners without having any customized postprocessing. The efficacy of the spin-echo echo planar imaging with view angle tilting technique in the correction of image distortion is demonstrated in phantom and in vivo brain imaging.

FIG. 1

The spin-echo EPI sequence with the VAT technique. The VAT gradient blips are applied along the SS direction at the same time as the PE gradient blips.

FIG. 2

Magnitude modulation functions for SE-EPI-VAT calculated from a RF slice profile used for in-vivo imaging. (a) θ =44.6° (b) θ =63.1°.

FIG. 3

Demonstration of image distortion along the PE direction (row in images) in SE-EPI in relation to parallel imaging and the VAT technique using the fat-air phantom. (a) SE-EP image shows both significant image distortion (black arrow) and chemical shift effects (arrow head). It also shows the considerable distortion of the phantom (white arrow). (b) Image from SE-EPI with GRAPPA (R=4) shows the reduction of distortion. However, it still has quite noticeable distortion (arrow) and chemical shift effects (arrow head) as well. (c) SE-EPI-VAT image shows severe image blurring although distortion correction can be appreciated. (d) Image from SE-EPI-VAT (θ =41.4°) with GRAPPA (R=4) shows distortion correction and the correction of chemical shift effects as well (arrows).

FIG. 4

Brain images of human frontal lobe. (a) SE image. (b) SE-EP image without fat suppression. The image shows significant image distortion near frontal lobe and the displacement of fatty tissue. (c) Image from SE-EPI with GRAPPA (R=4) without fat suppression. There still remain image distortion and chemical shift effects as well. (d) Image from SE-EPI-VAT (θ =48.9°) with GRAPPA (R=4). Both distortion and chemical shift effects are effectively corrected. Edges of brain region in the SE image were detected and overlaid onto each EP image for comparison.

FIG. 5

Human brain images of (a) frontal and (b) deep inferior OF region from an axial imaging. Edges of SE images are overlaid onto EP images to examine distortion and signal modulation effects. Regions of distortion and/or signal modulation effects are indicated by arrow(s) in R1 images. Parallel imaging (R2 and R4) has shown to reduce the artifacts due to the reduction of the effective echo spacing. However, when VAT was combined (R2-VAT and R4-VAT), the artifacts were corrected or further reduced significantly.

FIG. 6

Human brain images of (a) OF and (b) temporal region from a tilted-slice imaging (−20° from axial to coronal plane). Edges of SE images are overlaid onto EP images to examine distortion and signal modulation effects. Severe image distortion and/or signal modulation effects are seen in standard SE-EP images (R1). These artifacts are reduced by applying parallel imaging (R2 and R4) and further reduced or corrected by using VAT (R2-VAT and R4-VAT).

FIG. 7

(a) Representative saggital T1 image affine-registered to MNI space with the Frontal-pole region of Harvard-Oxford Cortical Structural Atlas overlaid in red color. (b) Bar graph shows an average number of voxels counted in the region shown in (a) calculated from 5 subjects with a standard deviation value indicated. (c) Difference of a voxel count (T1-EPI) was normalized by T1. From left to right in (b) and (c), results are presented for axial imaging and −20° and 20° (axial to coronal) tilted-slice imaging.