ISMRM 24th Annual Meeting & Exhibition • 07-13 May 2016 • Singapore

Scientific Session: New Frontiers in Image Reconstruction

Friday, May 13, 2016
Room 331-332
08:00 - 10:00
Moderators: Joseph Cheng, Justin Haldar

Learning a Variational Model for Compressed Sensing MRI Reconstruction
Kerstin Hammernik1, Florian Knoll2, Daniel K Sodickson2, and Thomas Pock1,3
1Institute for Computer Graphics and Vision, Graz University of Technology, Graz, Austria, 2Center for Biomedical Imaging and Center for Advanced Imaging Innovation and Research (CAI2R), Department of Radiology, NYU School of Medicine, New York, NY, United States, 3Safety & Security Department, AIT Austrian Institute of Technology GmbH, Vienna, Austria
Compressed sensing techniques allow MRI reconstruction from undersampled k-space data. However, most reconstruction methods suffer from high computational costs, selection of adequate regularizers and are limited to low acceleration factors for non-dynamic 2D imaging protocols. In this work, we propose a novel and efficient approach to overcome these limitations by learning a sequence of optimal regularizers that removes typical undersampling artifacts while keeping important details in the imaged objects and preserving the natural appearance of anatomical structures. We test our approach on patient data and show that we achieve superior results than commonly used reconstruction methods.

SENSE-LORAKS: Phase-Constrained Parallel MRI without Phase Calibration
Tae Hyung Kim1, Kawin Setsompop2, and Justin P. Haldar1
1Electrical Engineering, University of Southern California, Los Angeles, CA, United States, 2Radiology, Harvard Medical School, Boston, MA, United States
We introduce a novel framework called SENSE-LORAKS for partial Fourier phase-constrained parallel MRI reconstruction.  SENSE-LORAKS combines classical SENSE data modeling with advanced regularization based on the novel low-rank modeling of local k-space neighorhoods (LORAKS) framework.  Unlike conventional phase-constrained SENSE techniques, SENSE-LORAKS enables use of phase constraints without requiring a prior estimate of the image phase or a fully sampled region of k-space that could be used for phase autocalibration.  Compared to previous SENSE-based and LORAKS-based reconstruction approaches, SENSE-LORAKS is compatible with a much wider range of sampling trajectories, which can be leveraged to achieve much higher acceleration rates.

k-t ESPIRiT: Efficient Auto-Calibrated Parallel Imaging Reconstruction by Exploiting k-t Space Correlations
Claudio Santelli1, Adrian Huber1, and Sebastian Kozerke1
1Institute for Biomedical Engineering, University and ETH Zurich, Zurich, Switzerland
Using eigen-decomposition of a modified k-t SPIRiT operator, computationally optimized reconstruction formally translating into auto-calibrated SENSE-like reconstruction of a coil-combined x-f image (k-tESPIRiT) is proposed. 2D and 3D in-vivo experiments show equivalence of k-t SPIRiT and k-t ESPIRiT, and significant reconstruction time speed-up's of the proposed relative to the standard technique.

Motion-Resolved Golden-Angle Radial Sparse MRI Using Variable-Density Stack-of-Stars Sampling
Li Feng1, Tiejun Zhao2, Hersh Chandarana1, Daniel K Sodickson1, and Ricardo Otazo1
1Center for Advanced Imaging Innovation and Research (CAI2R), New York University School of Medicine, New York, NY, United States, 2Siemens Medical Solutions, New York, NY, United States
This work proposes a 3D free-breathing MRI technique called variable-density XD-GRASP, which employs stack-of-stars sampling with variable-density kz-undersampling and motion-resolved sparse reconstruction. The new sampling scheme combines the advantages of conventional stack-of-stars sampling and kooshball-type 3D radial sampling, enabling 3D continuous MRI with flexible slice resolution, robust fat suppression and low sensitivity to eddy currents. The performance of variable-density XD-GRASP is demonstrated for free-breathing liver MRI.

Non-Iterative Motion-Error Regularized Reconstruction for Efficient Respiratory Gating with Auto-Calibrating Parallel Imaging
Peng Lai1, Joseph Yitan Cheng2, Shreyas S Vasanawala2, and Anja C.S Brau3
1Global MR Applications and Workflow, GE Healthcare, Menlo Park, CA, United States, 2Radiology, Stanford University, Stanford, CA, United States, 3Global MR Applications and Workflow, GE Healthcare, Munich, Germany
Respiratory gating (RG) is commonly used for free-breathing 3D MRI. Conventional RG based on acceptance/rejection performs hard-threshholding on acquired data and suffers from either increased motion corruption with a large acceptance window or long scan time/increased undersampling artifacts with a small window. This work developed a non-iterative respiratory soft-threshholding method by incorporating the motion-induced error into autocalibrating parallel imaging (ac-PI). The proposed method showed more effective motion suppression on free-breathing 3D cine than conventional respiratory gating on the same datasets. This method can be generalized to suppress other types of motion with full acquisition or ac-PI as well.

Towards a Parameter-Free ESPIRiT: Soft-Weighting for Robust Coil Sensitivity Estimation
Siddharth Srinivasan Iyer1, Frank Ong1, and Michael Lustig1
1Electrical Engineering and Computer Sciences, University of California, Berkeley, Berkeley, CA, United States
ESPIRiT is a robust, auto-calibrating approach to parallel MR image reconstruction that estimates the subspace of sensitivity maps using an eigenvalue-based method. While it is robust to a range of parameter choices, having parameters that result in a tight subspace yields the best performance. We propose an automatic, parameter free method that appropriately weights the subspace using a shrinkage operator derived from Stein's Unbiased Risk Estimate. We demonstrate the efficacy of our technique by showing superior map estimation without user intervention in simulation and in-vivo data compared to the current default method of subspace estimation.

Self-Calibrated Phase-Correction for Superresolution of RASER at 7 T
Ute Goerke1
1CMRR/Radiology, University of Minnesota, Minneapolis, MN, United States
RASER (rapid acquisition with sequential excitation and refocusing) is an ultrafast imaging technique based on spatiotemporal encoding (SPEN). The excitation with a chirp-pulse with a low bandwidth-time product (R-value) introduces blurring in the SPEN dimension. Superresolution (SR) which removes the blurring fails as a result of the spatially varying B1-phase produced by radio-frequency coils at ultrahigh fields. A novel iterative phase-correction of the SR-algorithm is presented. It is shown that the spatial resolution and the SNR of blurred RASER images acquired at 7 T are significantly improved employing phase-corrected SR.

A convex source separation and reconstruction methodology for filtering dynamic contrast enhancement MRI data - Permission Withheld
Sudhanya Chatterjee1, Dattesh D Shanbhag1, Venkata Veerendranadh Chebrolu1, Uday Patil1, Sandeep N Gupta2, Moonjung Hwang 3, Jeong Hee Yoon4, Jeong Min Lee4, and Rakesh Mullick1
1GE Global Research, Bangalore, India, 2GE Global Research, Niskayuna, NY, United States, 3GE Healthcare, Seoul, Korea, Republic of, 4Seoul National University Hospital, Seoul, Korea, Republic of
Main aim of this research is to investigate a source separation based approach to remove noise from true signal, while maintaining original tissue enhancement signature. It is based on the hypothesis that there exists overlapping temporal information in the DCE-MRI data, which if identified, can be used for filtering noise out of the true concentration data. We demonstrate the utility of source separation and subsequent weight estimation methodology to filter “noise” from DCE concentration data and impact on the pK model parameters in liver DCE-MRI.

3D motion corrected SENSE reconstruction for multishot multislice MRI
Lucilio Cordero-Grande1, Emer Hughes1, Anthony Price1, Jana Hutter1, A. David Edwards1, and Joseph V. Hajnal1
1Centre for the Developing Brain, King's College London, London, United Kingdom
A framework for retrospectively motion corrected reconstruction of multislice multishot MRI in the presence of 3D rigid motion is developed. The method is able to cope both with within-plane and through-plane motion by estimating the motion states corresponding to the acquired shots and slices. It has been applied to 478 T1 and T2 newborn brain studies, including many severely motion corrupted examples, for which consistent structures have been recovered in more than 96% of cases. Due to its robustness and flexibility, our technique has wide potential application for both clinical and research examinations.

4D radial fat-suppressed alternating-TR bSSFP MRI with compressed sensing reconstruction for abdominal imaging during free breathing.
Jasper Schoormans1, Oliver Gurney-Champion1, Remy Klaassen2, Jurgen H. Runge1, Sonia I. Gonçalves3, Bram F. Coolen1, Abdallah G. Motaal1, Hanneke W.M. van Laarhoven2, Jaap Stoker1, Aart J. Nederveen1, and Gustav J. Strijkers4
1Department of Radiology, AMC, Amsterdam, Netherlands, 2Department of Medical Oncology, AMC, Amsterdam, Netherlands, 3Institute for Biomedical Imaging and Life Sciences, University of Coimbra, Coimbra, Portugal, 4Department of Biomedical Engineering and Physics, AMC, Amsterdam, Netherlands
We developed a 4D radial fat-suppressed alternating-TR bSSFP sequence with T2-like contrast for abdominal free-breathing imaging of pancreatic cancer patients. The sequence was tested in healthy volunteers and patients with pancreatic cancer and provided images of the abdomen during different respiratory motion states of diagnostic quality. 



The International Society for Magnetic Resonance in Medicine is accredited by the Accreditation Council for
Continuing Medical Education to provide continuing medical education for physicians.