Joint Annual Meeting ISMRM-ESMRMB 2014 10-16 May 2014 Milan, Italy

Novel Applications of Compressed Sensing

Tuesday 13 May 2014
Space 2  13:30 - 15:30 Moderators: Mariya Doneva, Ph.D., Justin P. Haldar, Ph.D.

13:30 0325.   
Accelerating MR Elastography with Sparse Sampling and Low-Rank Reconstruction
Curtis L Johnson1, Joseph L Holtrop1,2, Anthony G Christodoulou1,3, Matthew DJ McGarry4, John B Weaver4,5, Keith D Paulsen4,5, Zhi-Pei Liang1,3, John G Georgiadis1,6, and Bradley P Sutton1,2
1Beckman Institute, University of Illinois at Urbana-Champaign, Urbana, IL, United States, 2Bioengineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, 3Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, 4Thayer School of Engineering, Dartmouth College, Hanover, NH, United States, 5Dartmouth-Hitchcock Medical Center, Lebanon, NH, United States, 6Mechanical Science and Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States

Magnetic resonance elastography (MRE) requires the acquisition of a large number of images with differing gradient encoding direction, polarity, and displacement phase offsets. However, these images share a lot of information and can be represented through a reduced model order. In this work we demonstrate the ability to accelerate brain MRE acquisitions through sparse sampling and low-rank image reconstruction. Reducing the reconstructed model order from 48 to 10 resulted in virtually unchanged mechanical properties, and allowed for undersampling by factors up to 4x.

13:42 0326.   
Compressed Sensing 4D Flow Reconstruction using Divergence-Free Wavelet Transform
Frank Ong1, Martin Uecker1, Umar Tariq2, Albert Hsiao2, Marcus Alley2, Shreyas Vasanawala2, and Michael Lustig1
1Electrical Engineering and Computer Sciences, University of California, Berkeley, CA, United States, 2Radiology, Stanford University, CA, United States

In our previous work, divergence-free wavelet transform was shown to be effective in enforcing divergence-free constraints in denoising 4D flow data. In this work, we incorporate divergence-free wavelet in the compressed sensing iterative reconstruction process and present an accelerated 4D flow reconstruction method that is tolerant to phase wraps. Effects of phase wraps are reduced via phase cycle spinning, in which the phase is rotated randomly in each iteration, thereby preventing the need for phase unwrapping before reconstruction. The proposed reconstruction was applied on in-vivo data and was shown to yield better flow data from undersampled data that follow boundary conditions while maintaining core flow quantifications.

13:54 0327.   
Field-corrected imaging for sparsely-sampled fMRI by exploiting low-rank spatiotemporal structure
Hien Nguyen1 and Gary Glover2
1Department of Electronics & Computer Engineering, Hanoi University of Science & Technology, Hanoi, Vietnam, 2Department of Radiology, Stanford University, California, United States

Magnetic field gradients near air-tissue interfaces cause signal dropout, hampering BOLD fMRI. To make the data less prone to T2* susceptibility artifacts, it is desirable to reduce the readout duration. This can be achieved by undersampling k-space, which has been investigated for dynamic MRI and recently proposed for fMRI. In this work, we demonstrate a new field-corrected imaging approach to sparsely sampled fMRI, coined functional LOw Rank Approximations (fLORA). Specifically, we exploit partial separability (PS)-induced low rank structure of fMRI data via group-sparse regularization, combined with magnetic field inhomogeneity compensation.

14:06 0328.   Image quality characterization in Time-Resolved 3D CE-MRA
Yijing Wu1, Kevin M Johnson1, Jane H Maksimovic2, Charles A Mistretta1, and Patrick A Turski2
1Medical Physics, University of Wisconsin, Madison, WI, United States, 2Radiology, University of Wisconsin, Madison, WI, United States

Time-resolved 3D contrast-enhanced MR angiography (CE-MRA) often requires highly accelerated imaging to achieve clinically desired temporal and spatial resolutions. Recently developed non-linear reconstruction schemes, e.g. compressed sensing (CS) and HYPR, offer substantially greater acceleration than past methods but are unfortunately inherently object dependent and difficult to characterize with traditional linear methods. Subsequently, exemplary results in simplified digital phantoms do not translate clinically. In this work, we investigate the use of a highly realistic fractal based digital imaging phantom for accurate characterization of several non-linear reconstructions (CS, HYPR, and CS-HYPR).

14:18 0329.   
Towards Robust Breath-held 3D Abdominal DCE Imaging - permission withheld
Nadine Gdaniec1, Andrea J. Wiethoff2,3, Qing Yuan4, Peter Börnert5, Holger Eggers5, Daniella Pinho4, Ivan Pedrosa3,4, and Alfred Mertins1
1Institute for Signal Processing, University of Luebeck, Luebeck, Luebeck, Germany, 2Philips Research North America, Briarcliff Manor, New York, United States, 3Advanced Imaging Research Center, UT Southwestern Medical Center, Dallas, Texas, United States, 4Department of Radiology, UT Southwestern Medical Center, Dallas, Texas, United States, 5Philips Research Laboratories, Hamburg, Hamburg, Germany

Image quality in clinical dynamic contrast enhanced (DCE) MRI of the abdomen is often degraded by respiratory motion artifacts, making diagnosis difficult. Breath-holding is an efficient strategy to minimize respiration induced artifacts in the abdomen if the patient’s capability is sufficient. In many cases, patients have trouble holding their breath after contrast injection, even if they were able to do so earlier in the exam. To overcome this problem, DCE imaging with flexible scan termination is proposed in this work, to automatically adapt to the breath-hold capabilities of the patient. Shorter breath-holds are compromised with lower but adapted spatial resolution.

14:30 0330.   
Free Breathing Dynamic Contrast Enhanced 3D MRI with Resolved Respiratory Motion
Joseph Y. Cheng1,2, Tao Zhang1, John M. Pauly1, Shreyas S. Vasanawala2, and Michael Lustig3
1Electrical Engineering, Stanford University, Stanford, California, United States, 2Radiology, Stanford University, Stanford, California, United States,3Electrical Engineering & Computer Sciences, University of California, Berkeley, California, United States

Respiratory motion is a major hindrance in abdominal MRI. Previous work has been focused on reducing respiratory motion artifacts through correction schemes in the image reconstruction. An alternative approach is resolving the respiratory motion. We propose to correct for respiratory motion in free breathing DCE-MRI by extending the data-acquisition space to an additional respiratory dimension. The highly undersampled 5D-dataset is reconstructed by promoting the locally low-rank property in the DCE dimension and the total variation penalty in the respiratory motion dimension. The proposed technique achieves similar image quality to a DCE-only reconstruction and a lightly undersampled respiratory motion resolved reconstruction.

14:42 0331.   
Accelerated MPIO-Labeled Cell Imaging in the Heart
Anthony G. Christodoulou1, T. Kevin Hitchens2, Yijen L. Wu2, Zhi-Pei Liang1, and Chien Ho2
1Department of Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, 2Pittsburgh NMR Center for Biomedical Research, Department of Biological Sciences, Carnegie Mellon University, Pittsburgh, PA, United States

Low-rank (subspace) imaging with temporal navigation and sparse sampling of (k, t)-space has previously been used to accelerate several cardiac imaging applications. Here we describe a more efficient self-navigated pulse sequence to acquire both navigator and sparse data in a single TR, doubling imaging speed to approach 100 frames per second. We demonstrate the assessment of myocardial inflammation in rats through self-navigated T2*-weighted imaging of immune cells labeled with micron-sized paramagnetic iron oxide (MPIO) particles.

14:54 0332.   Fast 3D Free-breathing Abdominal Dynamic Contrast Enhanced MRI with High Spatiotemporal Resolution
Tao Zhang1, Joseph Cheng1,2, Marcus Alley2, Martin Uecker3, Michael Lustig3, John Pauly1, and Shreyas Vasanawala2
1Electrical Engineering, Stanford University, Stanford, California, United States, 2Radiology, Stanford University, Stanford, California, United States,3Electrical Engineering and Computer Sciences, UC Berkeley, Berkeley, California, United States

Dynamic Contrast Enhanced (DCE) MRI is commonly used to detect and characterize lesions. A free-breathing DCE acquisition has high scan efficiency, but image quality can be compromised by respiratory motion. In this work, a soft-gated locally low rank parallel imaging reconstruction method is proposed for highly accelerated 3D free-breathing DCE MRI. The proposed method can significantly reduce motion artifacts, and provide high spatiotemporal resolution (approximately 1 mm3 spatial resolution and 4 s frame rate). The proposed method has been validated on in vivo datasets.

Jingyuan Lyu1, Pascal Spincemaille2, Martin R Prince2, Yi Wang2, Fuquan Ren1, and Leslie Ying1
1Department of Biomedical Engineering, Department of Electrical Engineering, The State University of New York at Buffalo, Buffalo, New York, United States, 2Weill Cornell Medical College, New York, New York, United States

This abstract presents a novel method to effectively integrate spiral acquisition, parallel imaging, partial separable (PS) model, and sparsity constraints for highly accelerated dynamic contrast enhanced MRI. In data acquisition, a phased array coil was used to continuously acquire data along a stack of variable-density spirals updated with the golden angle. In reconstruction, with the sparsity constraints, the coil sensitivities, spatial and temporal bases of the PS model are jointly estimated through alternating optimization. Experimental results from in vivo DCE liver imaging data demonstrate the proposed method is able to achieve both high spatial and temporal resolution.

15:18 0334.   High angularly resolved diffusion imaging with accelerated multi-shot acquisition and compressed sensing
Tzu-Cheng Chao1,2, Jr-Yuan George Chiou3, Stephan E. Maier3, and Bruno Madore3
1Department of Computer Science and Information Engineering, National Cheng-Kung University, Tainan, Taiwan, 2Institute of Medical Informatics, Naitonal Cheng-Kung University, Tainan, Taiwan, 3Department of Radiology, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts, United States

High angularly resolved diffusion imaging is a well-established strategy to help resolve fiber crossings and enable tractography. Long scan times and the presence of geometrical distortion in the resulting images may be the main factors currently limiting its clinical use. In the present work, methods are combined for accelerating the acquisition process in k-space (to reduce distortion) as well as in the diffusion-encoding space (to reduce scan time). As compared to a non-accelerated protocol, results are presented that offer a four-fold reduction in distortion as well as a reduction by about 40% in scan time.