3965

Blip-Up Blip-Down Correction for Head MR Elastography
Roger C Grimm1, Kevin J Glaser2, and Richard L Ehman2

1Radiology, Mayo Clinic, Rochester, MN, United States, 2Mayo Clinic, Rochester, MN, United States

Synopsis

Geometric distortion in echo planar imaging-spin echo (EPI-SE) images caused by a non-uniform B0 field remains an issue in clinical imaging. In elastography, this geometric distortion leads to systematic errors in stiffness estimates. While blip-up blip-down correction is routinely used in diffusion imaging in the head, this correction is not used in MR elastography. The purpose of this work is to assess the effect of applying distortion correction on elastography scans and the resulting stiffness estimates. We show that the corrections provide reduced variability in stiffness maps with no additional acquisition time required.

Purpose

Geometric distortion in echo planar imaging-spin echo (EPI-SE) images caused by a non-uniform B0 field remains an issue in clinical imaging. In MR elastography (MRE), geometric distortions in phase difference images distort the apparent wavelength and introduce bias into the resulting stiffness values. While blip-up blip-down correction is routinely used in diffusion imaging in the head1-3, this correction is not used in MRE. The purpose of this work is to apply the blip-up blip-down correction in head MRE scans and assess the effect on the resulting stiffness estimates.

Methods

Studies were performed with Institutional Review Board approved protocols. Scans were obtained with a MRE single shot EPI-SE based pulse sequence on a 3.0T HDxt GE scanner. Sampling parameters for the sequence were: acquired resolution 96 x 96, TR = 4.0s, TE = 78.8ms, FOV = 24cm, slices = 48, slice thickness = 3mm, ASSET reduction = 2, mechanical frequency = 60Hz, phases=8, and scan time=6:28. Two complete scans were performed with the ky traversal direction reversed to provide a distortion pair of blip-up and blip-down images. A synthetic data set was derived from a subset of these two scans with processing shown in Figure 1. The correction scheme employs the routines topup and applytopup3. To minimize smoothing effects, Jacobian interpolation was used with applytopup.

Results

Scans of a head mimicking PVC phantom4 are shown in Figure 2. The opposing geometric distortions in the magnitude and curl images are seen in the blip-up and blip-down images. In areas of spread-out distortion the apparent wavelength increases and the resulting stiffness estimates increase. Likewise, in areas of signal pile-up, the reverse is true and a decrease in the stiffness estimate is seen. The images using topup correction show reduced distortions in the magnitude and curl images. Figure 3 shows that the uncorrected images have 3 to 6% error relative to the stiffness results of the corrected images. The results from a scan of a normal volunteer are seen in Figure 4 and they also show good distortion correction in both the magnitude and curl images. The distortion corrected stiffness estimates for the in-vivo scan are generally between the blip-up and blip-down stiffness values.

Discussion

The inherent ability of the blip-up blip-down correction approach to resolve the signal pile-up ambiguity makes it a good candidate to apply to an application that already uses multiple time samples to increase the signal to noise ratio. As long as a minimum of four phase samples are acquired in an MRE scan, the blip-up blip-down correction can be used without an increase in scan time. The results show that without correction, areas of signal spread-out and pile-up will have systematic errors in the stiffness estimates.

Future directions would include comparing separately obtained field maps with those derived from the topup algorithm and looking into the interpolation methods used for correction that might be better suited to the phase information used in elastography processing.

Acknowledgements

NIH EB001981

References

1. Chang H, Fitzpatrick JM. A technique for accurate magnetic resonance imaging in the presence of field inhomogeneities. IEEE Trans Med Imag 1992;11:319–329. 2. M. Okan Irfanoglu, Pooja Modi, Amritha Nayak, Elizabeth B. Hutchinson, Joelle Sarlls, and Carlo Pierpaoli. DR-BUDDI (Diffeomorphic Registration for Blip-Up blip-Down Diffusion Imaging) Method for Correcting Echo Planar Imaging Distortions. Neuroimage. 2015 Feb 1; 106: 284–299. 3. Andersson JL, Skare S, Ashburner J. How to correct susceptibility distortions in spin-echo echo-planar images: application to diffusion tensor imaging. Neuroimage 2003;20:870–888. 1367 4. Arvin Arani, et al. High Resolution 3D GRE Brain MR Elastography is Feasible with a High Performance Compact 3T Scanner, ISMRM proceedings, 2016, 1367.

Figures

Figure 1 Diagram of the processing pipeline. To assess the effect of topup processing, complete scans were performed for both blip-up and blip-down data acquisition orders. A synthetic scan data set was obtained by using a subset of four phase samples from of each of the blip-up and blip-down scans. This provides the same amount of signal and the same total scan time for the three comparisons. To generate the phase difference images needed for stiffness estimation, the real and imaginary image sets are processed separately then combined after correction to provide the phase information.

Figure 2 Axial slice through the cerebellum portion of a PVC head phantom. Note the spread out distortion indicated by the arrows. Also note the over-estimation in the stiffness maps in the same location. The topup corrected processing recovers and corrects much of the distortion in both the magnitude and curl images. The distortion correction allows for a more uniform stiffness estimate.

Figure 3 In areas of spread-out distortion, the apparent wavelength increases and the resulting stiffness estimates increase. Likewise, in areas of signal pile-up, the reverse is true and a decrease in the stiffness estimate is seen. Two regions of interest (ROI) from the slice in Figure 2 show this effect.

Figure 4 Axial slice of a healthy volunteer showing opposing areas of “pile up” and “spread out” distortion. As in the phantom example, topup processing recovers and corrects much of the distortion in both the magnitude and curl images. The stiffness estimate difference is greatest between the blip-up and blip-down scans indicating that topup processing can reduce variability caused by geometric distortion.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
3965