(C-102) Acceleration of Magnetic Resonance Elastography using a Novel Distributed Encoding Technique

Introduction:: The mechanical properties of tissues are indicative of underlying health and as such have been used in many diagnostic capacities. Magnetic resonance elastography (MRE) is a non-invasive, in vivo, MRI technique which provides maps of mechanical properties such as stiffness by imaging the displacement of tissue in the brain in response to mechanical actuation^1,2. These MRE-derived mechanical properties can sensitively provide insight into microstructural changes in the brain that differentially occur with aging and disease progression³. Major constraints of this technique are that long scan times are necessary to encode three directions of motion across time, which limits clinical adoption and makes it susceptible to subject movement. We propose a novel sampling technique and estimation scheme called EDGE (Elastography with Distributed, Generalized Encoding) that distributes encodings of direction and time across samples to reduce the number of samples needed to estimate complex, full vector displacement fields. This algorithm provides additional opportunity for acceleration through EDGE’s novel implementation of non-traditional sampling directions in an optimized encoding matrix to further increase sampling efficiency. Sampling reduction through EDGE takes a two-pronged approach to MRE data quality issues by minimizing opportunity for subject motion to occur through reduction of scan time and by retrospectively rejecting images disrupted by any unwanted motion that does occur.


Materials and Methods:: Traditional MRE separately samples displacement along three orthogonal directions with four or eight phase offsets to capture wave propagation through tissue with time, resulting in a total of 24 to 48 images^2,4. The total sampling is in excess of the theoretical minimum – seven, accounting for real and imaginary components of three directions of motion, plus background phase – because of how each direction is sampled separately in time and processed via Fourier transforms to extract harmonic motion fields. We have generated an encoding matrix (Fig. 1) that does not adhere to the traditional sampling pattern, but encodes along an optimized set of directions that are distributed uniformly across the surface of a sphere, as is common in diffusion tensor imaging⁵. This allows every sample to include information from three axes, thus contributing information relevant to estimating each displacement in each direction. Eschewing the traditional Fourier technique, we formulate the estimation of motion from these samples as an optimization problem using a general encoding matrix and minimizing the objective function, following analogous examples in field mapping by Fessler^6,7. We tested the feasibility and performance of EDGE through retrospective sampling of realistic simulated brain data using the proposed EDGE sampling scheme. We used a set of 24 phase images, mimicking the total number of images typically acquired in traditional MRE, and reducing down to the theoretical minimum of seven⁸. Stiffness maps were generated using a nonlinear inversion algorithm⁹ and compared with traditionally sampled and processed MRE data using normalized-root-mean-square error (NRMSE).


Results, Conclusions, and Discussions:: Using simulated data, we demonstrated that the proposed EDGE sampling and estimation technique can accurately recover original displacement fields and stiffness maps with high quality and agreement with traditional MRE even after high levels of undersampling (Fig. 2). The EDGE estimation algorithm is fast, taking less than 3 minutes to reconstruct a full 3D dataset on a desktop computer. This technique provides accurate reconstruction of stiffness maps, with NRMSE of 10.3% and average error of 6.2% using the theoretical minimum sampling which provides 3.4x acceleration. Some error is seen even in the fully sampled case (NRMSE = 6.5% and average error = 3.8%), the disagreement of the two stiffness maps is mostly localized to the boundaries of the brain tissue, likely due to our current roughness penalty used for regularization in the EDGE algorithm. To further improve agreement between stiffness maps generated with EDGE vs traditionally collected MRE data, we plan to introduce a number of improvements to the estimation technique, including performing optimization of a regularization term to stabilize the results and minimize the effects of noise. Additional parameter optimization can be accomplished through optimization of the encoding matrix so that sampling of unique directions and time points can be done with maximum efficiency. Next steps will include implementation of EDGE in current MRE sequences to collect data with reduced sampling in human subjects, allowing us to circumvent issues from subject motion in MRE acquisition and data quality. Successful implementation of this technique in MRE sequences will enable prospective acceleration or retrospective volume rejection – or a combination of both – for uncooperative participants, significantly advancing the adoptability of MRE.


Acknowledgements (Optional): : National Institutes of Health: R01-EB027577

References (Optional): : 1. Muthupillai, R. et al. Magnetic resonance elastography by direct visualization of propagating acoustic strain waves. Science (80-. ). 269, 1854–1857 (1995).

2. Manduca, A. et al. Magnetic resonance elastography: Non-invasive mapping of tissue elasticity. Med. Image Anal. 5, 237–254 (2001).

3. Hiscox, L. V., Schwarb, H., McGarry, M. D. J. & Johnson, C. L. Aging brain mechanics: Progress and promise of magnetic resonance elastography. Neuroimage 232, (2021).

4. Nir, G., Sahebjavaher, R. S., Sinkus, R. & Salcudean, S. E. A framework for optimization-based design of motion encoding in magnetic resonance elastography. Magn. Reson. Med. 73, 1514–1525 (2015).

5. Jones, D. K., Horsfield, M. A. & Simmons, A. Optimal strategies for measuring diffusion in anisotropic systems by magnetic resonance imaging. Magn. Reson. Med. 42, 515–525 (1999).

6. Funai, A. K., Fessler, J. A., Yeo, D. T. B., Noll, D. C. & Olafsson, V. T. Regularized field map estimation in MRI. IEEE Trans. Med. Imaging 27, 1484–1494 (2008).

7. Lin, C. Y. & Fessler, J. A. Efficient Regularized Field Map Estimation in 3D MRI. IEEE Trans. Comput. Imaging 6, 1451–1458 (2020).

8. Trzasko, J. et al. Theoretical Performance and Sampling Limits in Steady-State Magnetic Resonance Elastography. in International Society of Magnetic Resonance in Medicine 2518 (2015).

9. McGarry, M. D. J. et al. Multiresolution MR elastography using nonlinear inversion. Med. Phys. 39, 6388–6396 (2012).