Removal of EPI Nyquist ghost artifacts with two‐dimensional phase correction

Abstract

Odd-even echo inconsistencies result in Nyquist ghost artifacts in the reconstructed EPI images. The ghost artifacts reduce the image signal-to-noise ratio and make it difficult to correctly interpret the EPI data. In this article a new 2D phase mapping protocol and a postprocessing algorithm are presented for an effective Nyquist ghost artifacts removal. After an appropriate k-space data regrouping, a 2D map accurately encoding lowand high-order phase errors is derived from two phase-encoded reference scans, which were originally proposed by Hu Keywords:echo planar imaging; EPI; Nyquist ghost artifactPhase inconsistencies between EPI echoes corresponding to different readout gradient polarities generate Nyquist ghost artifacts in the reconstructed images. The Nyquist ghost artifacts not only degrade EPI image quality, but also result in inaccurate EPI based quantitative measurements, such as EPI-based apparent diffusion coefficient (ADC) mapping (1). These phase inconsistencies may originate from different factors, such as field inhomogeneities (2), eddy current effects (3,4), imperfect gradient waveforms (5), and anisotropic gradient delays in oblique-plane EPI (6).Three types of approaches have been reported to date to remove EPI ghost artifacts. The first approach is 1D phase correction based on information from reference scans. The phase shift information can be measured with either a single non-phase encoded reference scan (7) or two phaseencoded scans (2). The reference information is then applied to correct phase errors using 1D Fourier transform modulation in subsequently obtained EPI data. This approach is easy to implement for both single-shot and segmented EPI. The phase errors along the phase-encoding direction, however, cannot be corrected with this 1D approach. In oblique plane EPI, for example, 2D phase errors due to anisotropic gradient delays cannot be completely removed with 1D correction (6). In addition, spatially dependent eddy current effects could also generate highorder echo shifts along both readout and phase-encoding directions. The second approach is 2D correction with pulse sequence compensation (4,6). It has been shown that, using a modified EPI pulse sequence with additional compensation blips, 2D linear phase errors due to hardware imperfections can be reduced (4,6). Since no reference image is required, this method is appropriate for EPI based real-time scans, such as cardiac imaging. However, high-order phase errors due to spatially dependent eddy current effects may not be effectively removed with this approach. In addition, its tolerance to subject-dependent field inhomogeneities has not been investigated. The third approach removes ghost artifacts by estimating phase errors from the EPI data itself. As described by Buonocore and Gao (5), an accurate 2D map indicating odd-even echo inconsistencies can be calculated from odd-echo-only and even-echo-only EPI data, provided that the parent image and the ghost do not overlap in the original EPI. In this case, high-order phase err...

Snapshot gradient-recalled echo-planar images of rat brains at long echo time at 9.4 T

“…The remaining ghosting artifacts could be explained by uneven gradient performance with different polarities, suggested by other studies [25,26]. However, this would not change our observation that, other than the improved gradient performance, adjusting B 0 field inhomogeneities at high fields plays a very important role in obtaining highquality GRE-EPI images with high spatial resolution at long TE.…”

High‐resolution fMRI with higher‐order generalized series imaging and parallel imaging techniques (HGS‐parallel)

“…However, the images from the EPI sequence suffer from artifacts such as Nyquist ghosts and severe geometric distortions, which should be reduced by additional postprocessing. Several ways of correcting these artifacts have been proposed, including the acquisition of separate reference images (33), the modification of EPI sequences (34), and algorithmic correction schemes (35).…”

Method for spatially interleaving two images to halve EPI readout times: Two reduced acquisitions interleaved (TRAIL)

“…4 was used. A nonuniform Nyquist ghost, which resulted from offresonance effects, was removed by the method of Chen and Wyrwicz (29). Additionally, a map of the magnetic field was measured, and distortion in the EPI image was corrected (26) by use of the approach described by Jezzard and Balaban (30).…”