Chaoyi Zhang¹, Yihang Zhou¹, Jingyuan Lyu¹, Ukash Nakarmi¹, and Leslie Ying^1,2
1Electrical Engineering, State University at buffalo,SUNY, Buffalo, NY, United States, ²Biomedical Engineering, State University at Buffalo,SUNY, Buffalo, NY, United States

MR parameter mapping has shown great potential but is still limited in clinical application due to the lengthy acquisition time. To address this issue, we  proposed a novel image reconstruction method(PARMA) to accelerate parameter mapping with reduced multi-channel acquisition using alternating projections on the single-exponential parametric manifold, the subspace data consistancy, and the convex of the regularized coil sensitivities. The experimental results show the potential of highly accelerated quantitative imaging by the proposed method.

Anthony G. Christodoulou^1,2, Jaime L. Shaw^2,3, Behzad Sharif^2,4, and Debiao Li^2,3
1Heart Institute, Cedars-Sinai Medical Center, Los Angeles, CA, United States, ²Biomedical Imaging Research Institute, Cedars-Sinai Medical Center, Los Angeles, CA, United States, ³Department of Bioengineering, University of California, Los Angeles, Los Angeles, CA, United States, ⁴Department of Biomedical Sciences, Cedars-Sinai Medical Center, Los Angeles, CA, United States

We present a general low-rank tensor framework for high-dimensional cardiac imaging, modeling the underlying image as partially separable in all relevant dimensions: space, cardiac phase, respiratory phase, wall-clock time (e.g., for contrast agent dynamics), variable sequence parameters (e.g., inversion time), etc. An explicit-subspace variant of the framework is demonstrated, with subspaces estimated from navigator data and a signal recovery dictionary of solutions to the Bloch equations (similar to MR fingerprinting). This variant is used to perform ECG-less cardiac- and time-resolved T₁ mapping during first-pass perfusion, as well as free-breathing, ECG-less native T₁ mapping at multiple cardiac phases. The framework shows promise for time-resolved T₁ mapping and other high-dimensional applications. 

Yi Guo¹, Sajan Goud Lingala¹, Yinghua Zhu¹, R. Marc Lebel², and Krishna S Nayak^1
1Electrical Engineering, University of Southern California, Los Angeles, CA, United States, ²GE Healthcare, Calgary, AB, Canada

Pharmacokinetic (PK) parameter maps derived from DCE-MRI provide quantitative physiological information that aids in cancer diagnosis and assessment of treatment response. Recently, direct reconstruction of PK maps from under-sampled k,t-space has shown great potential to provide optimal detection of kinetic parameter maps from an information theoretic perspective. We build on prior work (using the Patlak model) and demonstrate direct reconstruction of kinetic parameter maps using the extended-Tofts model, which is a more appropriate model in brain tumor. We demonstrate convergence behavior, computational efficiency, and application to brain DCE-MRI.

Itthi Chatnuntawech¹, Patrick McDaniel¹, Stephen F. Cauley^2,3, Borjan A. Gagoski^3,4, Christian Langkammer⁵, Adrian Martin⁶, Ellen Grant^3,4, Lawrence L. Wald^2,3,7, Kawin Setsompop^2,3, Elfar Adalsteinsson^1,7,8, and Berkin Bilgic^2
1Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, MA, United States, ²Department of Radiology, A. A. Martinos Center for Biomedical Imaging, Charlestown, MA, United States, ³Harvard Medical School, Boston, MA, United States, ⁴Fetal-Neonatal Neuroimaging & Developmental Science Center, Boston Children’s Hospital, Boston, MA, United States, ⁵Department of Neurology, Medical University of Graz, Graz, Austria, ⁶Applied Mathematics, Universidad Rey Juan Carlos, Madrid, Spain, ⁷Harvard-MIT Health Sciences and Technology, Cambridge, MA, United States, ⁸Institute for Medical Engineering and Science, Cambridge, MA, United States

To directly estimate tissue magnetic susceptibility distribution from the raw phase of a gradient echo acquisition, we propose a single-step quantitative susceptibility mapping (QSM) method that benefits from its three components: (i) the single-step processing that prevents error propagation normally encountered in multiple-step QSM algorithms, (ii) multiple spherical mean value kernels that permit high fidelity background removal, and (iii) total generalized variation regularization that promotes a piecewise-smooth solution without staircasing artifacts. A fast solver for the proposed method, which enables simple analytical solutions for all of the optimization steps, is also developed. Improved image quality over conventional QSM algorithms is demonstrated using the SNR-efficient Wave-CAIPI and 3D-EPI acquisitions.

Dongwook Lee¹, Kyong Hwan Jin¹, Eung-yeop Kim², Sunghong Park¹, and Jong Chul Ye^1
1Bio and Brain Engineering, Korea Advanced Institute of Science and Technology, Daejeon, Korea, Republic of, ²Radiology, Gachon University Gil Medical Center, Inchoen, Korea, Republic of

The purpose of this study is to develop an accelerated MR parameter mapping technique. For accelerated T1 and T2 mapping, spin-echo inversion recovery and multi-echo spin echo pulse sequences were redesigned to perform undersampling along phase encoding direction. The highly missing k-space were then interpolated by using recently proposed annihilating filter based low-rank Hankel matrix approach (ALOHA). By exploiting the duality between the transform domain sparsity and the low-rankness of weighted Hankel structured matrix in k-space, ALOHA provided outperforming reconstruction results compared to the existing compressed sensing methods.

Bo Zhao¹, Kawin Setsompop¹, Borjan Gagoski², Huihui Ye¹, Elfar Adalsteinsson³, P. Ellen Grant², and Larry L. Wald^1
1Athinoula A. Martinos Center for Biomedical Imaging, Chalestown, MA, United States, ²Boston Children's Hospitial, Boston, MA, United States, ³EECS, MIT, Cambridge, MA, United States

A new model-based approach using low-rank and sparsity constraints is presented for reconstructing the accelerated magnetic resonance fingerprinting (MRF) time-series images. By enabling high-quality reconstructions of contrast-weighted images from highly-undersampled data,  the proposed method produces more accurate estimates of tissue parameter maps compared to the conventional gridding based reconstruction of the time-series. Ultimately, the goal is to reduce imaging time for MRF acquisitions and improve spatial resolution.  

Haifeng Wang¹, Emre Kopanoglu¹, R. Todd Constable^1,2, and Gigi Galiana^1
1Department of Radiology and Biomedical Imaging, Yale University, New Haven, CT, United States, ²Department of Neurosurgery, Yale University, New Haven, CT, United States

Low-Rank O-Space presents a scheme to incorporate O-Space imaging with Low-Rank matrix recovery. The Low-Rank reconstruction based on iterative nonlinear conjugate gradient algorithm is applied to substitute the previous Kaczmarz and Compressed Sensing (CS) reconstructions to recover highly undersampled O-Space data. The simulations and experiments illustrate the proposed scheme can remove artifacts and noise in O-Space imaging at high reduction factors, compared to results recovered by Kaczmarz and CS. Moreover, the proposed method does not need to modify the conventional O-Space pulse sequences, and reconstruction results are better than those in radial imaging recovered by Kaczmarz, CS, or Low-Rank methods.

Florian Knoll¹, Martin Holler², Thomas Koesters¹, Martijn Cloos¹, Ricardo Otazo¹, Kristian Bredies², and Daniel K Sodickson^1
1Center for Advanced Imaging Innovation and Research (CAI2R) and Bernard and Irene Schwartz Center for Biomedical Imaging, Department of Radiology, NYU School of Medicine, New York, NY, United States,²Mathematics and Scientific Computing, University of Graz, Graz, Austria

A typical clinical imaging protocol covers a large number of different image contrasts and, in the era of multi-modality systems, even different imaging modalities. While the resulting datasets share a substantial amount of structural information, they consist of fundamentally different contrasts and signal values and show unique features and image content. We propose a reconstruction framework based on nuclear-norm second-order Total Generalized Variation that exploits structural similarity both between different contrasts and modalities while still being flexible with respect to signal intensity and unique features. Numerical simulations and in vivo MR-Fingerprinting experiments demonstrate improved PET resolution and improved depiction of quantitative values. The proposed approach allows a 6 minute whole brain coverage exam that provides both quantitative PET and MR-relaxation parameters.

Maojing Fu¹, Jonghye Woo², Marissa Barlaz³, Ryan Shosted³, Zhi-Pei Liang¹, and Bradley Sutton^4
1Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ²CAMIS (Center for Advanced Medical Imaging Sciences), Massachusetts General Hospital, Boston, MA, United States, ³Linguistics, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ⁴Bioengineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States

Dynamic speech MRI holds great promise for visualizing articulatory motion in the vocal tract. Recent work has enabled accelerated imaging speed, resulting in the need to integrate mechanisms to enable interpretation of the dynamic images that contain great amounts of movement information. This work integrates a spatiotemporal atlas into a partial separable (PS) model-based imaging framework and uses the atlas as prior information to improve reconstruction quality. This method not only captures high-quality dynamics at 102 frames per second, but also enables quantitative characterization of articulatory variability utilizing the residual component from the atlas-based sparsity constraint.

Chao Ma¹, Fan Lam¹, Qiang Ning^1,2, Bryan A. Clifford^1,2, Qiegen Liu¹, Curtis L. Johnson¹, and Zhi-Pei Liang^1,2
1Beckman Institute, University of Illinois Urbana-Champaign, Urbana, IL, United States, ²Electrical and Computer Engineering, University of Illinois Urbana-Champaign, Urbana, IL, United States

Dynamic MRSI measures the temporal changes of metabolite concentrations by acquiring a time series of MRSI data. These data can be used in a range of applications, including the study of the response of a metabolic system to a perturbation. However, high-resolution dynamic MRSI is challenging due to poor SNR resulting from the low concentrations of metabolites. This work presents a new method for high-resolution dynamic 31P-MRSI using high-order partially separable functions. The method has been validated using in vivo dynamic 31P-MRSI experiments, producing encouraging results.