View Presentation Video

Moritz Blumenthal^1,2, H. Christian M. Holme², and Martin Uecker^1,2
1Institute for Diagnostic and Interventional Radiology, University Medical Center Göttingen, Göttingen, Germany, ²Institute of Biomedical Imaging, Graz University of Technology, Graz, Austria

Keywords: Image Reconstruction, Software Tools

We propose a new decomposition of the nuFFT algorithm, allowing for a zero-memory-overhead implementation of the (adjoint-)nuFFT. With one grid-sized buffer, the decomposition allows for a memory efficient nuFFT with negligible computational overhead compared to a two-fold oversampled conventional nuFFT - to the prize of less efficient parallelization on many-threads CPU systems. We reduce memory requirements of 3D non-Cartesian PICS-reconstructions in BART by a factor up to ten, allowing for GPU acceleration of reconstructions with eight coils and matrix size 256x256x256 on a 4GB consumer-level GPU.

View Presentation Video

Sijie Zhong^1,2, Minjia Chen³, Ke Dai^1,4, Hao Chen^1,2, Lucio Frydman⁴, and Zhiyong Zhang^1,2
1School of Biomedical Engineering, Shanghai Jiao Tong University, Shanghai, China, ²Institute of Medical Robotics, Shanghai Jiao Tong University, Shanghai, China, ³Department of Engineering, University of Cambridge, Cambridgeshire, United Kingdom, ⁴Department of Chemical and Biological Physics, Weizmann Institute of Science, Rehovot, Israel

Keywords: Image Reconstruction, Artifacts, SPEN

The reconstruction theory of Spatio-temporal encoding MRI is not complete at present. Aliasing artifacts in SPEN MRI reconstructions can be traced to image contributions corresponding to high-frequency k-space signals. The k-space picture provides the spatial displacements, phase offsets and linear amplitude modulations associated to these artifacts, as well as routes to removing these from the reconstruction results. These new ways to estimate the artifact priors were applied to reduce SPEN reconstruction artifacts on simulated, phantom and human brain MRI data.

View Presentation Video

Riwei Jin¹, Yudu Li², Fangxu Xing³, Imani Gilbert⁴, Jamie Perry⁴, Jonghye Woo³, Zhi-pei Liang⁵, and Brad Sutton^1
1Department of Bioengineering, University of Illinois Urbana-Champaign, Champaign, IL, United States, ²National Center for Supercomputing Applications, Champaign, IL, United States, ³Gordon Center for Medical Imaging, Department of Radiology, Massachusetts General Hospital/Harvard Medical School, Boston, MA, United States, ⁴Department of Communication Sciences and Disorders, East Carolina University, Greenville, NC, United States, ⁵6 Department of Electrical and Computer Engineering, University of Illinois Urbana-Champaign, Champaign, IL, United States

Keywords: Image Reconstruction, Sparse & Low-Rank Models

To optimize the reconstruction quality of isotropic 3D dynamic speech magnetic resonance imaging with large scan volume, we applied two novel methods based on the Partial Separability model theory: 1. Locally High-Rank reconstruction through Partial Separability model (PS-LHR) which enables higher rank to be devoted to the dynamic speech region. 2. Implementation of Regional-Optimized Temporal basis (ROT) to focus the temporal navigator information on the speech region. The improvement in reconstruction quality was seen to decrease the noise of regions of the image outside the area of interest and increase dynamic smoothness in the speech region.

View Presentation Video

Daniel I Gendin¹ and Jiaen Liu^2
1Advanced Imaging Research Center, UT Southwestern Medical Center, Dallas, TX, United States, ²Advanced Imaging Research Center and Radiology, UT Southwestern Medical Center, Dallas, TX, United States

Keywords: Image Reconstruction, Brain

Head motion can lead to artifacts in MR images. One source of motion artifacts is dynamic nonlinear variation in the background magnetic field (B₀). Fully accounting for this variation is computationally intractable. We proposed an efficient method of performing this correction by making use of principal component analysis and considering only a few dominant modes. We applied our method to measured data and compared the results to the case with no correction as well correction using a previous K-means based method.

Dongbiao Sun^1,2, Yan Zhuo^1,2, Lin Chen^2,3, and Zihao Zhang^1,3
1State Key Laboratory of Brain and Cognitive Science, Beijing MR Center for Brain Research, Institute of Biophysics, Chinese Academy of Sciences, Beijing, China, ²University of Chinese Academy of Sciences, Beijing, China, ³Institute of Artificial Intelligence, Hefei Comprehensive National Science Center, Hefei, China

Keywords: Image Reconstruction, Image Reconstruction, Non-Fourier transform reconstruction

Modern MRI reconstructs images by performing Fourier transform (FT) on k-space to decompose the signal on an orthogonal basis composed of trigonometric functions. We are inspired by the problem of estimating the Direction of Arrival in radar theory and propose a novel route for MRI reconstruction based on multiple signal classification (MUSIC). MUSIC-MRI overcomes the sidelobe problem of FT and significantly promotes the actual resolution in the meaning of full width at half maximum (FWHM). Our phantom experiments show that the FWHMs are 0.45mm by MUSIC-MRI and 1.50mm by FT, while the nominal resolution of the k-space data is 0.94mm. 

View Presentation Video

Xinzeng Wang¹, Daniel Litwiller², Arnaud Guidon³, Tim Sprenger⁴, and Robert Marc Lebel^5
1GE Healthcare, Houston, TX, United States, ²GE Healthcare, Denver, CO, United States, ³GE Healthcare, Boston, MA, United States, ⁴GE Healthcare, Stockholm, Sweden, ⁵GE Healthcare, Calgary, AB, Canada

Keywords: Image Reconstruction, Artifacts, Partial Fourier

A partial Fourier acquisition has been widely used for fast MR imaging. To reduce the truncation artifacts in partial Fourier image, Homodyne reconstruction is often used, and it exploits the conjugate symmetry in real-valued signal to recover the full k-space. However, the MR signal is complex-valued. Artifacts are commonly observed in Homodyne images in the regions of rapid phase change due to the interference of imaginary components of adjacent pixels. In this work, we proposed a modified Homodyne reconstruction to reduce the conventional Homodyne artifacts and truncation artifacts by using a high-resolution phase from a pre-trained deep-learning network. 

View Presentation Video

Tao Hong¹, Jeffrey A. Fessler², and Luis Hernandez-Garcia ^1
1Department of Radiology, University of Michigan, Ann Arbor, MI, United States, ²Department of Electrical and Computer Engineering, University of Michigan, Ann Arbor, MI, United States

Keywords: Image Reconstruction, Data Processing, Fast reconstruction algorithm

This work studies a complex quasi-Newton proximal method (CQNPM) for MRI reconstruction using wavelets or total variation (TV) based regularization. Our experiments show that our method is faster than the accelerated proximal method [1,2] in terms of iteration and CPU time.  

View Presentation Video

Paul I. Dubovan*^1,2, Gabriel Varela-Mattatall*^1,2, Tales Santini^1,2, Kyle M. Gilbert^1,2, Ravi S. Menon^1,2, and Corey A. Baron^1,2
1Centre for Functional and Metabolic Mapping (CFMM), Robarts Research Institute, Western University, London, ON, Canada, ²Department of Medical Biophysics, Schulich School of Medicine and Dentistry, Western University, London, ON, Canada

Keywords: Image Reconstruction, Image Reconstruction

Reconstruction problems involving the expanded encoding model and field monitoring are typically solved using least squares optimization with the conjugate gradient method and early stopping as an implicit form of regularization. However, this is likely a suboptimal strategy for low SNR acquisitions, such as accelerated or high-resolution diffusion MRI. Hence, in this work we present an extension to the matMRI reconstruction toolbox that incorporates compressed sensing regularization. Results demonstrate that the expanded encoding model and compressed sensing regularization are complementary tools that mitigate artifacts from phase perturbations while permitting lower SNR conditions for high-resolution single-shot spiral diffusion-weighted imaging. 

View Presentation Video

Soumick Chatterjee^1,2,3, Philipp Ernst^1,2, Oliver Speck^3,4,5, and Andreas Nürnberger^1,2,5
1Faculty of Computer Science, Otto von Guericke University Magdeburg, Magdeburg, Germany, ²Data and Knowledge Engineering Group, Otto von Guericke University Magdeburg, Magdeburg, Germany, ³Department of Biomedical Magnetic Resonance, Otto von Guericke University Magdeburg, Magdeburg, Germany, ⁴German Center for Neurodegenerative Disease, Magdeburg, Germany, ⁵Center for Behavioral Brain Sciences, Magdeburg, Germany

Keywords: Image Reconstruction, Artifacts

Iterative undersampled MRI reconstructions, such as compressed sensing, can reconstruct undersampled MRIs - but due to their slow execution speed, they are not suitable for real-time applications. Several deep learning approaches have been proposed, mostly working in image space. Some of the approaches, which work on the k-space or in a mix of spaces, employ real-valued convolutions splitting the complex k-space into real and imaginary parts for processing - destroying the geometric relationship within the data. This research proposes Fourier-PD and Fourier-PDUNet models using complex-valued convolutions, which attempt to predict missing k-space frequencies and also to reduce artefacts in the image space. 

View Presentation Video

Qing-San Xiang^1
1Radiology, University of British Columbia, Vancouver, BC, Canada

Keywords: Image Reconstruction, Image Reconstruction, water-fat imaging

Water-Fat Imaging with only a single acquisition is desirable for its pulse sequence simplicity and scan time efficiency.  However, robust reconstruction is more challenging due to limited available information.  In this work, it is discovered that some previously unused information can be found by statistical analysis, leading to effective reconstruction of water and fat images.  In particular, Independent Component Analysis (ICA) was used to determine the phase error.  Well separated water and fat images were subsequently obtained after phase correction.

Jie Yang¹, Yujian liu^1,2, Wenjia Hang¹, Meining Chen³, Jianqi Li⁴, Yinqiao Yi⁴, Haodong Zhong⁴, and Xu Yan^3
1Department of Radiology, Zigong First People's Hospital, Zigong, China, ²Sichuan Vocational College of Health and Rehabilitation, Zigong, China, ³MR Scientific Marketing, Siemens Healthineers, Shanghai, China, ⁴Shanghai Key Laboratory of Magnetic Resonance, East China Normal University, Shanghai, China

Keywords: Image Reconstruction, MSK

Quantitative imaging of intervertebral disc degeneration is important for the diagnosis of lower back pain. In our study, a simulTaneous multi-relaXation-time Imaging (TXI) method was evaluated, which generated bone marrow fat fraction (PDFF), R2* and R1 mapping within a single scan. The results showed that the BMFF, T2*, T1 mapping calculated from TXI method were close to the values reported in other studies for lumbar discs.

View Presentation Video

Jeehun Kim^1,2, Zhiyuan Zhang^1,3, Ruiying Liu⁴, Brendan Eck¹, Mingrui Yang¹, Hongyu Li⁴, Mei Li¹, Richard Lartey¹, Carl S. Winalski^1,5, Leslie Ying^4,6, and Xiaojuan Li^1,5
1Department of Biomedical Engineering, Program of Advanced Musculoskeletal Imaging (PAMI), Cleveland Clinic, Cleveland, OH, United States, ²Department of Electrical, Computer, and Systems Engineering, Case Western Reserve University, Cleveland, OH, United States, ³Department of Biomedical Engineering, Case Western Reserve University, Cleveland, OH, United States, ⁴Electrical Engineering, University at Buffalo, State University of New York, Buffalo, NY, United States, ⁵Department of Diagnostic Radiology, Imaging Institute, Cleveland Clinic, Cleveland, OH, United States, ⁶Biomedical Engineering, University at Buffalo, State University of New York, Buffalo, NY, United States

Keywords: Image Reconstruction, Cartilage

Quantitative MR T1rho T2 imaging shows promising results on detecting early-stage osteoarthritis, but long scan time limits the spatial resolution, making it vulnerable to partial volume averaging. Such effect reduces the sensitivity to small focal degeneration. In this study, compressed sensing reconstruction with spatio-temporal finite difference regularization was used to accelerate high-resolution (slice thickness < 2mm) T1rho imaging and standard-resolution simultaneous T1rho and T2 imaging, and evaluated the result comparing with reference imaging, retrospective and prospective reconstruction, and scan-rescan repeatability. 

View Presentation Video

Nikolai J Mickevicius¹, Andrew S Nencka¹, and Kevin M Koch^1
1Medical College of Wisconsin, Milwaukee, WI, United States

Keywords: Image Reconstruction, Image Reconstruction

In multi-spectral imaging near metallic implants, redundant information is present in neighboring spectral bins and contrasts due to the overlapping and smooth spectral profiles. This redundancy is exploited here to reduce scan time when multiple image contrasts are desired.

View Presentation Video

Xiaoqing Wang^1,2, Jaejin Cho^1,2, Yohan Jun^1,2, Borjan Gagoski³, and Berkin Bilgic^1,2
1Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Charlestown, MA, United States, ²Department of Radiology, Harvard Medical School, Boston, MA, United States, ³Fetal-Neonatal Neuroimaging and Developmental Science Center, Boston Children's Hospital, Harvard Medical School, Boston, MA, United States

Keywords: Image Reconstruction, Brain, model-based recontruction

Phase-based T2 mapping acquires two GRE volumes with opposite RF phase increments, then computes the phase difference and relates this to T2 values through Bloch modeling. We exploit the redundancy in these two acquisitions by posing this as a nonlinear inverse problem, and directly estimate the phase difference from k-space. This reduces the problem to the estimation of a complex M0 volume, and a real-valued phase difference, thus reducing the number of real-valued unknowns from 4 to 3 per voxel. The proposed reconstruction flexibly admits additional regularization, and yields higher quality images from undersampled acquisitions for rapid T2 mapping.

View Presentation Video

Nolan Meyer^1,2, Myung-Ho In¹, David F Black¹, Norbert G Campeau¹, Kirk M Welker¹, John Huston III¹, Matt A Bernstein¹, and Joshua D Trzasko^1
1Radiology, Mayo Clinic, Rochester, MN, United States, ²Mayo Clinic Graduate School of Biomedical Sciences, Rochester, MN, United States

Keywords: Image Reconstruction, Neuro, Clinical, radiologist, evaluation

A model-based image reconstruction (MBIR) platform was developed for accelerated point spread function encoded echo planar imaging. Beginning with retrospectively truncated data, images were reconstructed with the enhanced MBIR framework. These were compared by a team of neuroradiologists with images obtained using standard reconstructions from non-truncated datasets. Scores were assigned for evaluation criteria including high-contrast resolution (HCR), low-contrast detectability (LCD), signal to noise ratio (SNR), artifacts, and overall preference. Nonparametric statistical test results indicate preference for MBIR in HCR, LCD, and overall preference, with comparable results for SNR and artifacts. MBIR thus facilitates PSF-EPI acceleration, enhancing clinical practicality.

View Presentation Video

Charles Iglehart¹, Ali Bilgin^1,2,3, and Manojkumar Saranathan^4,5
1Electrical and Computer Engineering, The University of Arizona, Tucson, AZ, United States, ²Biomedical Engineering, The University of Arizona, Tucson, AZ, United States, ³Medical Imaging, The University of Arizona, Tucson, AZ, United States, ⁴Radiology, UMass Chan Medical School, Worcester, MA, United States, ⁵Neuroscience, Morningside Graduate School of Biomedical Sciences, Worcester, MA, United States

Keywords: Image Reconstruction, Brain

$$$T_{2}^{*}$$$ and quantitative susceptibility mapping are of increasing clinical interest, but typically require lengthy ME-GRE scans. To accelerate such acquisitions, we propose and implement a volumetric reconstruction method employing locally low rank (LLR) techniques to recover complex-valued ME images from highly undersampled k-space data that is suitable for parameter mapping. We present results for reconstructed magnitude and phase across multiple datasets and acceleration ratios, as well as computed $$$R_{2}^{*}$$$ and QSM from reconstructed images.

View Presentation Video

Eva S Peper^1,2, Adèle LC Mackowiak^1,2,3, Berk C Açıkgöz^1,2, Nils Plähn^1,2, Yasaman Safarkhanlo⁴, Li Feng⁵, and Jessica AM Bastiaansen^1,2
1Department of Diagnostic, Interventional and Pediatric Radiology (DIPR), Inselspital, Bern University Hospital, University of Bern, Bern, Switzerland, ²Translation Imaging Center (TIC), Swiss Institute for Translational and Entrepreneurial Medicine, Bern, Switzerland, ³Department of Diagnostic and Interventional Radiology, Lausanne University Hospital (CHUV) and University of Lausanne (UNIL), Lausanne, Switzerland, ⁴Department of Cardiology, University Hospital Bern, Bern, Switzerland, ⁵BioMedical Engineering and Imaging Institute (BMEII), Department of Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, United States

Keywords: Image Reconstruction, Sparse & Low-Rank Models, bSSFP, phase-cycled bSSFP, subspace reconstruction

Fat fraction (FF) maps can be estimated from phase-cycled (PC) balanced steady-state free precession (bSSFP) imaging using a dictionary fitting approach (SPARCQ). Three-dimensional PC bSSFP imaging, however, is time consuming, since around 16 PCs must be sampled to fully describe a voxels’ PC profile. Compressed sensing (CS) has been proven to accelerate MRI scans. This study has two aims 1) to evaluate the performance of two CS algorithms (total variation (TV)- vs. subspace-constrained) taking regularisation along the PC dimension into account and 2) to investigate the upper limits of acceleration using these techniques on FF quantification using SPARCQ.

View Presentation Video

Paul Kokeny¹, Qiuyun Xu¹, Sara Gharabaghi¹, Sean Sethi^1,2, Yu Liu³, Youmin Zhang³, Peng Liu³, Naying He³, Fuhua Yan³, and E. Mark Haacke^1,2,4
1SpinTech MRI, Bingham Farms, MI, United States, ²Radiology, Wayne State University, Detroit, MI, United States, ³Shanghai Jiao Tong University School of Medicine, Ruijin Hospital, Shanghai, China, ⁴Neurology, Wayne State University, Detroit, MI, United States

Keywords: Image Reconstruction, Quantitative Imaging

By using the inherent relationship between proton spin density (PSD) and T1, we propose a new image-processing approach to reduce noise called CROWN (Constrained Reconstruction of White Noise). Firstly, we established a linear relationship between these two parameters, then applied a cost function to constrain simulated Strategically Acquired Gradient Echo (STAGE) PSD map and T1 map data in the presence of noise. Secondly, we applied this approach to in vivo STAGE images to reduce noise and improve SNR without the loss of image detail. CROWN has the potential to make higher resolution or faster imaging viable with improved SNR. 

View Presentation Video

Nitin Jain¹, Ashok Kumar P Reddy¹, Rajdeep Das¹, Sajith Rajamani¹, Rajagopalan Sundaresan¹, Harsh Kumar Agarwal¹, M Ramasubba Reddy², and Ramesh Venkatesan^1
1GE Healthcare, Bangalore, India, ²Indian Institute of Technology Madras, Chennai, India

Keywords: Image Reconstruction, Body

Diffusion weighted MR imaging (DWI) is key to pathology detection in anatomies such as brain, abdomen and prostate. Echo planar imaging (EPI) provides a rapid means to acquire DWI. EPI with variable k-space sampling scheme and an auto-calibrating image reconstruction technique, vARC, has recently been shown to reduce distortion in DWI and improve the image quality in single channel volume coil/body coil acquisitions. Here, we propose a new low rank reconstruction technique for robust reconstruction and improved image quality for DWI acquired using vARC’s EPI multi-shot acquisition scheme with single channel body coil.

View Presentation Video

Yan Dai¹, Jie Deng¹, and Xun Jia^2
1University of Texas Southwestern Medical Center, Dallas, TX, United States, ²Johns Hopkins University, Baltimore, MD, United States

Keywords: Image Reconstruction, Quantitative Imaging

We developed a joint quantitative imaging reconstruction algorithm to reconstruct T2-weighted (T2W) images and T2 mapping simultaneously through an iterative optimization process, which integrated the weighted linear regression fitting. The proposed method achieved accurate T2 measurements and improved signal-to-noise (SNR) of both T2W images and the T2 map.