We proposed a motion-correction method for joint reconstruction of blip-up/down EPI acquisition (BUDA-EPI) of brain diffusion MRI. Motion parameters were estimated and incorporated into the joint parallel imaging reconstruction of the blip-up/down multi-shot data, which included B0 field maps and Hankel structured low-rank constraint. The proposed motion-corrected reconstruction approach was demonstrated in vivo to provide motion-robust reconstruction of blip-up/down multi-shot EPI diffusion data.
1. Andersson JL, Skare S, Ashburner J. How to correct susceptibility distortions in spin-echo echo-planar images: application to diffusion tensor imaging. NeuroImage, 20(2), 870-888.
2. Liao C, Cao X, Cho J, Zhang Z, Setsompop K, Bilgic B. Highly efficient MRI through multi-shot echo planar imaging. Wavelets and Sparsity XVIII 11138, 2019.
3. Brown TT, Kuperman JM, Erhart M, White NS, Roddey JC, Shankaranarayanan A, Han ET, Rettmann D, Dale AM. Prospective motion correction of high-resolution magnetic resonance imaging data in children. Neuroimage 2010;53(1):139-145.
4. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: sensitivity encoding for fast MRI. Magnetic Resonance in Medicine, 1999; 42(5), 952-962.
5. Uecker M, Lai P, Murphy MJ, Virtue P, Elad M, Pauly JM, et al. ESPIRiT--an eigenvalue approach to autocalibrating parallel MRI: where SENSE meets GRAPPA. Magn Reson Med. 2014;71(3):990-1001.
6. Low-Rank Modeling of Local k-Space Neighborhoods (LORAKS) for Constrained MRI. IEEE Transactions on Medical Imaging, 2014; 33(3): 668-681.
7. Mani M, Jacob M, Kelley D, Magnotta V. Multi-shot sensitivity-encoded diffusion data recovery using structured low-rank matrix completion (MUSSELS). Magn Reson Med. 2017;78(2):494-507.
8. Shin PJ, Larson PE, Ohliger MA, Elad M, Pauly JM, Vigneron DB, et al. Calibrationless parallel imaging reconstruction based on structured low-rank matrix completion. Magn Reson Med. 2014;72(4):959-70.
9.Jenkinson M, Bannister P, Brady JM and Smith SM. Improved Optimisation for the Robust and Accurate Linear Registration and Motion Correction of Brain Images. NeuroImage, 17(2), 825-841, 2002.
Figure 1.
(A) Flowchart of BUDA reconstruction. A Hankel structured low-rank constrained forward model is used to jointly reconstruct distortion-free images.
(B) A corrupted field map was estimated from TOPUP when motion occurred between blip-up/down shots, resulting in a corrupted image reconstruction.
(C) Flowchart of MOCO-BUDA reconstruction. The base field map E0 was transformed to fit position of shot, and images xt were corrected to the same base position across shots to enable the Hankel structured low-rank model, resulting in motion-corrected distortion-free images.
Figure 2.
(A) The flowchart of motion estimation based on MCFLIRT. Images from all shots are reconstructed separately with SENSE, and transformation parameters are found using MCFLIRT between shots. The transformation parameters are used to update the B0 map position, which is then used to correct the distortion. Updated images are used to find new transformation parameters, continuing iteratively to improve motion estimates.
(B) Using a transformed field map, the difference is much closer to the actual field map at POSn and the correlation between maps is close to unity.
Figure 3.
The reference no-motion acquisition (left), motion acquisition without correction (middle) and with motion-correction (right) images are compared with different levels of motion. For mild motion (top), the image without correction exhibited blurring and slight artifacts, while for moderate motion (middle), more distortion artifacts occurred. The artifacts in both cases could be recovered using the proposed method. When motion was significant (bottom), artifacts are severe, and the motion-corrected image is improved, yet still suffers from slight blurring.
Figure 4.
(A) Time-varying motion estimation of rotation and x/y-axis translation during a 24-direction DWI acquisition.
(B) Reconstructed images with/without motion correction of 4 respective time frames indicated by red boxes in (A).
(C) The averaged DWI of the motion-free period (directions 1-10) and moving period (directions 11~24).
Figure 5.
Two slices in three orthorhombic views of the 1.5-mm isotropic resolution 3D images with/without motion correction. The 6-dimension motion parameters are shown in the top for both blip-up and –down shots.