scispace - formally typeset
Search or ask a question

Showing papers in "Magnetic Resonance in Medicine in 2018"


Journal ArticleDOI
TL;DR: In this paper, a variational network approach is proposed to reconstruct the clinical knee imaging protocol for different acceleration factors and sampling patterns using retrospectively and prospectively undersampled data.
Abstract: PURPOSE To allow fast and high-quality reconstruction of clinical accelerated multi-coil MR data by learning a variational network that combines the mathematical structure of variational models with deep learning. THEORY AND METHODS Generalized compressed sensing reconstruction formulated as a variational model is embedded in an unrolled gradient descent scheme. All parameters of this formulation, including the prior model defined by filter kernels and activation functions as well as the data term weights, are learned during an offline training procedure. The learned model can then be applied online to previously unseen data. RESULTS The variational network approach is evaluated on a clinical knee imaging protocol for different acceleration factors and sampling patterns using retrospectively and prospectively undersampled data. The variational network reconstructions outperform standard reconstruction algorithms, verified by quantitative error measures and a clinical reader study for regular sampling and acceleration factor 4. CONCLUSION Variational network reconstructions preserve the natural appearance of MR images as well as pathologies that were not included in the training data set. Due to its high computational performance, that is, reconstruction time of 193 ms on a single graphics card, and the omission of parameter tuning once the network is trained, this new approach to image reconstruction can easily be integrated into clinical workflow. Magn Reson Med 79:3055-3071, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

1,111 citations


Journal ArticleDOI
TL;DR: To demonstrate accurate MR image reconstruction from undersampled k‐space data using cross‐domain convolutional neural networks (CNNs) using cross-domain Convolutional Neural Networks, a parallel version of TSP, is presented.
Abstract: Purpose To demonstrate accurate MR image reconstruction from undersampled k-space data using cross-domain convolutional neural networks (CNNs) METHODS: Cross-domain CNNs consist of 3 components: (1) a deep CNN operating on the k-space (KCNN), (2) a deep CNN operating on an image domain (ICNN), and (3) an interleaved data consistency operations. These components are alternately applied, and each CNN is trained to minimize the loss between the reconstructed and corresponding fully sampled k-spaces. The final reconstructed image is obtained by forward-propagating the undersampled k-space data through the entire network. Results Performances of K-net (KCNN with inverse Fourier transform), I-net (ICNN with interleaved data consistency), and various combinations of the 2 different networks were tested. The test results indicated that K-net and I-net have different advantages/disadvantages in terms of tissue-structure restoration. Consequently, the combination of K-net and I-net is superior to single-domain CNNs. Three MR data sets, the T2 fluid-attenuated inversion recovery (T2 FLAIR) set from the Alzheimer's Disease Neuroimaging Initiative and 2 data sets acquired at our local institute (T2 FLAIR and T1 weighted), were used to evaluate the performance of 7 conventional reconstruction algorithms and the proposed cross-domain CNNs, which hereafter is referred to as KIKI-net. KIKI-net outperforms conventional algorithms with mean improvements of 2.29 dB in peak SNR and 0.031 in structure similarity. Conclusion KIKI-net exhibits superior performance over state-of-the-art conventional algorithms in terms of restoring tissue structures and removing aliasing artifacts. The results demonstrate that KIKI-net is applicable up to a reduction factor of 3 to 4 based on variable-density Cartesian undersampling.

323 citations


Journal ArticleDOI
TL;DR: A new fully automated musculoskeletal tissue segmentation method using deep convolutional neural network (CNN) and three‐dimensional (3D) simplex deformable modeling to improve the accuracy and efficiency of cartilage and bone segmentation within the knee joint.
Abstract: Purpose To describe and evaluate a new fully automated musculoskeletal tissue segmentation method using deep convolutional neural network (CNN) and three-dimensional (3D) simplex deformable modeling to improve the accuracy and efficiency of cartilage and bone segmentation within the knee joint. Methods A fully automated segmentation pipeline was built by combining a semantic segmentation CNN and 3D simplex deformable modeling. A CNN technique called SegNet was applied as the core of the segmentation method to perform high resolution pixel-wise multi-class tissue classification. The 3D simplex deformable modeling refined the output from SegNet to preserve the overall shape and maintain a desirable smooth surface for musculoskeletal structure. The fully automated segmentation method was tested using a publicly available knee image data set to compare with currently used state-of-the-art segmentation methods. The fully automated method was also evaluated on two different data sets, which include morphological and quantitative MR images with different tissue contrasts. Results The proposed fully automated segmentation method provided good segmentation performance with segmentation accuracy superior to most of state-of-the-art methods in the publicly available knee image data set. The method also demonstrated versatile segmentation performance on both morphological and quantitative musculoskeletal MR images with different tissue contrasts and spatial resolutions. Conclusion The study demonstrates that the combined CNN and 3D deformable modeling approach is useful for performing rapid and accurate cartilage and bone segmentation within the knee joint. The CNN has promising potential applications in musculoskeletal imaging. Magn Reson Med, 2017. © 2017 International Society for Magnetic Resonance in Medicine.

247 citations


Journal ArticleDOI
TL;DR: To develop a super‐resolution technique using convolutional neural networks for generating thin‐slice knee MR images from thicker input slices, and compare this method with alternative through‐plane interpolation methods.
Abstract: PURPOSE To develop a super-resolution technique using convolutional neural networks for generating thin-slice knee MR images from thicker input slices, and compare this method with alternative through-plane interpolation methods. METHODS We implemented a 3D convolutional neural network entitled DeepResolve to learn residual-based transformations between high-resolution thin-slice images and lower-resolution thick-slice images at the same center locations. DeepResolve was trained using 124 double echo in steady-state (DESS) data sets with 0.7-mm slice thickness and tested on 17 patients. Ground-truth images were compared with DeepResolve, clinically used tricubic interpolation, and Fourier interpolation methods, along with state-of-the-art single-image sparse-coding super-resolution. Comparisons were performed using structural similarity, peak SNR, and RMS error image quality metrics for a multitude of thin-slice downsampling factors. Two musculoskeletal radiologists ranked the 3 data sets and reviewed the diagnostic quality of the DeepResolve, tricubic interpolation, and ground-truth images for sharpness, contrast, artifacts, SNR, and overall diagnostic quality. Mann-Whitney U tests evaluated differences among the quantitative image metrics, reader scores, and rankings. Cohen's Kappa (κ) evaluated interreader reliability. RESULTS DeepResolve had significantly better structural similarity, peak SNR, and RMS error than tricubic interpolation, Fourier interpolation, and sparse-coding super-resolution for all downsampling factors (p < .05, except 4 × and 8 × sparse-coding super-resolution downsampling factors). In the reader study, DeepResolve significantly outperformed (p < .01) tricubic interpolation in all image quality categories and overall image ranking. Both readers had substantial scoring agreement (κ = 0.73). CONCLUSION DeepResolve was capable of resolving high-resolution thin-slice knee MRI from lower-resolution thicker slices, achieving superior quantitative and qualitative diagnostic performance to both conventionally used and state-of-the-art methods.

243 citations


Journal ArticleDOI
TL;DR: A novel fast method for reconstruction of multi‐dimensional MR fingerprinting (MRF) data using deep learning methods and it is shown that this method can be used to solve the challenge of integrating 3D image recognition and 3D handwriting analysis.
Abstract: Demonstrate a novel fast method for reconstruction of multi-dimensional MR fingerprinting (MRF) data using deep learning methods.A neural network (NN) is defined using the TensorFlow framework and trained on simulated MRF data computed with the extended phase graph formalism. The NN reconstruction accuracy for noiseless and noisy data is compared to conventional MRF template matching as a function of training data size and is quantified in simulated numerical brain phantom data and International Society for Magnetic Resonance in Medicine/National Institute of Standards and Technology phantom data measured on 1.5T and 3T scanners with an optimized MRF EPI and MRF fast imaging with steady state precession (FISP) sequences with spiral readout. The utility of the method is demonstrated in a healthy subject in vivo at 1.5T.Network training required 10 to 74 minutes; once trained, data reconstruction required approximately 10 ms for the MRF EPI and 76 ms for the MRF FISP sequence. Reconstruction of simulated, noiseless brain data using the NN resulted in a RMS error (RMSE) of 2.6 ms for T1 and 1.9 ms for T2 . The reconstruction error in the presence of noise was less than 10% for both T1 and T2 for SNR greater than 25 dB. Phantom measurements yielded good agreement (R2 = 0.99/0.99 for MRF EPI T1 /T2 and 0.94/0.98 for MRF FISP T1 /T2 ) between the T1 and T2 estimated by the NN and reference values from the International Society for Magnetic Resonance in Medicine/National Institute of Standards and Technology phantom.Reconstruction of MRF data with a NN is accurate, 300- to 5000-fold faster, and more robust to noise and dictionary undersampling than conventional MRF dictionary-matching.

242 citations


Journal ArticleDOI
TL;DR: A novel deep learning approach with domain adaptation is proposed to restore high‐resolution MR images from under‐sampled k‐space data to solve the problem of streaking artifact patterns in magnetic resonance imaging.

241 citations


Journal ArticleDOI
TL;DR: To develop an efficient acquisition for high‐resolution diffusion imaging and allow in vivo whole‐brain acquisitions at 600‐ to 700‐μm isotropic resolution.
Abstract: Purpose To develop an efficient acquisition for high-resolution diffusion imaging and allow in vivo whole-brain acquisitions at 600- to 700-μm isotropic resolution. Methods We combine blipped-controlled aliasing in parallel imaging simultaneous multislice (SMS) with a novel slab radiofrequency (RF) encoding gSlider (generalized slice-dithered enhanced resolution) to form a signal-to-noise ratio–efficient volumetric simultaneous multislab acquisition. Here, multiple thin slabs are acquired simultaneously with controlled aliasing, and unaliased with parallel imaging. To achieve high resolution in the slice direction, the slab is volumetrically encoded using RF encoding with a scheme similar to Hadamard encoding. However, with gSlider, the RF-encoding bases are specifically designed to be highly independent and provide high image signal-to-noise ratio in each slab acquisition to enable self-navigation of the diffusion's phase corruption. Finally, the method is combined with zoomed imaging (while retaining whole-brain coverage) to facilitate low-distortion single-shot in-plane encoding with echo-planar imaging at high resolution. Results A 10-slices-per-shot gSlider-SMS acquisition was used to acquire whole-brain data at 660 and 760 μm isotropic resolution with b-values of 1500 and 1800 s/mm2, respectively. Data were acquired on the Connectome 3 Tesla scanner with 64-channel head coil. High-quality data with excellent contrast were achieved at these resolutions, which enable the visualization of fine-scale structures. Conclusions The gSlider-SMS approach provides a new, efficient way to acquire high-resolution diffusion data. Magn Reson Med 79:141–151, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

152 citations


Journal ArticleDOI
TL;DR: The aim of the 2016 quantitative susceptibility mapping (QSM) reconstruction challenge was to test the ability of various QSM algorithms to recover the underlying susceptibility from phase data faithfully.
Abstract: Purpose The aim of the 2016 quantitative susceptibility mapping (QSM) reconstruction challenge was to test the ability of various QSM algorithms to recover the underlying susceptibility from phase data faithfully. Methods Gradient-echo images of a healthy volunteer acquired at 3T in a single orientation with 1.06 mm isotropic resolution. A reference susceptibility map was provided, which was computed using the susceptibility tensor imaging algorithm on data acquired at 12 head orientations. Susceptibility maps calculated from the single orientation data were compared against the reference susceptibility map. Deviations were quantified using the following metrics: root mean squared error (RMSE), structure similarity index (SSIM), high-frequency error norm (HFEN), and the error in selected white and gray matter regions. Results Twenty-seven submissions were evaluated. Most of the best scoring approaches estimated the spatial frequency content in the ill-conditioned domain of the dipole kernel using compressed sensing strategies. The top 10 maps in each category had similar error metrics but substantially different visual appearance. Conclusion Because QSM algorithms were optimized to minimize error metrics, the resulting susceptibility maps suffered from over-smoothing and conspicuity loss in fine features such as vessels. As such, the challenge highlighted the need for better numerical image quality criteria. Magn Reson Med, 2017. © 2017 International Society for Magnetic Resonance in Medicine.

152 citations


Journal ArticleDOI
TL;DR: A new segmentation method using deep convolutional neural network, 3D fully connected conditional random field, and 3D simplex deformable modeling to improve the efficiency and accuracy of knee joint tissue segmentation is described and evaluated.
Abstract: Purpose To describe and evaluate a new segmentation method using deep convolutional neural network (CNN), 3D fully connected conditional random field (CRF), and 3D simplex deformable modeling to improve the efficiency and accuracy of knee joint tissue segmentation. Methods A segmentation pipeline was built by combining a semantic segmentation CNN, 3D fully connected CRF, and 3D simplex deformable modeling. A convolutional encoder-decoder network was designed as the core of the segmentation method to perform high resolution pixel-wise multi-class tissue classification for 12 different joint structures. The 3D fully connected CRF was applied to regularize contextual relationship among voxels within the same tissue class and between different classes. The 3D simplex deformable modeling refined the output from 3D CRF to preserve the overall shape and maintain a desirable smooth surface for joint structures. The method was evaluated on 3D fast spin-echo (3D-FSE) MR image data sets. Quantitative morphological metrics were used to evaluate the accuracy and robustness of the method in comparison to the ground truth data. Results The proposed segmentation method provided good performance for segmenting all knee joint structures. There were 4 tissue types with high mean Dice coefficient above 0.9 including the femur, tibia, muscle, and other non-specified tissues. There were 7 tissue types with mean Dice coefficient between 0.8 and 0.9 including the femoral cartilage, tibial cartilage, patella, patellar cartilage, meniscus, quadriceps and patellar tendon, and infrapatellar fat pad. There was 1 tissue type with mean Dice coefficient between 0.7 and 0.8 for joint effusion and Baker's cyst. Most musculoskeletal tissues had a mean value of average symmetric surface distance below 1 mm. Conclusion The combined CNN, 3D fully connected CRF, and 3D deformable modeling approach was well-suited for performing rapid and accurate comprehensive tissue segmentation of the knee joint. The deep learning-based segmentation method has promising potential applications in musculoskeletal imaging.

143 citations


Journal ArticleDOI
TL;DR: This article introduces a constrained imaging method based on low‐rank and subspace modeling to improve the accuracy and speed of MR fingerprinting (MRF).
Abstract: Purpose This article introduces a constrained imaging method based on low-rank and subspace modeling to improve the accuracy and speed of MR fingerprinting (MRF). Theory and methods A new model-based imaging method is developed for MRF to reconstruct high-quality time-series images and accurate tissue parameter maps (e.g., T1 , T2 , and spin density maps). Specifically, the proposed method exploits low-rank approximations of MRF time-series images, and further enforces temporal subspace constraints to capture magnetization dynamics. This allows the time-series image reconstruction problem to be formulated as a simple linear least-squares problem, which enables efficient computation. After image reconstruction, tissue parameter maps are estimated via dictionary-based pattern matching, as in the conventional approach. Results The effectiveness of the proposed method was evaluated with in vivo experiments. Compared with the conventional MRF reconstruction, the proposed method reconstructs time-series images with significantly reduced aliasing artifacts and noise contamination. Although the conventional approach exhibits some robustness to these corruptions, the improved time-series image reconstruction in turn provides more accurate tissue parameter maps. The improvement is pronounced especially when the acquisition time becomes short. Conclusions The proposed method significantly improves the accuracy of MRF, and also reduces data acquisition time. Magn Reson Med 79:933-942, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

140 citations


Journal ArticleDOI
TL;DR: To develop a quantitative susceptibility mapping (QSM) method with a consistent zero reference using minimal variation in cerebrospinal fluid (CSF) susceptibility, data are analyzed through a probabilistic approach.
Abstract: Purpose To develop a quantitative susceptibility mapping (QSM) method with a consistent zero reference using minimal variation in cerebrospinal fluid (CSF) susceptibility. Theory and methods The ventricular CSF was automatically segmented on the R2* map. An L2 -regularization was used to enforce CSF susceptibility homogeneity within the segmented region, with the averaged CSF susceptibility as the zero reference. This regularization for CSF homogeneity was added to the model used in a prior QSM method (morphology enabled dipole inversion [MEDI]). Therefore, the proposed method was referred to as MEDI+0 and compared with MEDI in a numerical simulation, in multiple sclerosis (MS) lesions, and in a reproducibility study in healthy subjects. Results In both the numerical simulations and in vivo experiments, MEDI+0 not only decreased the susceptibility variation within the ventricular CSF, but also suppressed the artifact near the lateral ventricles. In the simulation, MEDI+0 also provided more accurate quantification compared to MEDI in the globus pallidus, substantia nigra, corpus callosum, and internal capsule. MEDI+0 measurements of MS lesion susceptibility were in good agreement with those obtained by MEDI. Finally, both MEDI+0 and MEDI showed good and similar intrasubject reproducibility. Conclusion QSM with a minimal variation in ventricular CSF is viable to provide a consistent zero reference while improving image quality. Magn Reson Med 79:2795-2803, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Journal ArticleDOI
TL;DR: The purpose of this study was to develop and implement experimental strategies for using [1‐13C]pyruvate to probe in vivo metabolism for patients with brain tumors and other neurological diseases.
Abstract: Author(s): Park, Ilwoo; Larson, Peder EZ; Gordon, Jeremy W; Carvajal, Lucas; Chen, Hsin-Yu; Bok, Robert; Van Criekinge, Mark; Ferrone, Marcus; Slater, James B; Xu, Duan; Kurhanewicz, John; Vigneron, Daniel B; Chang, Susan; Nelson, Sarah J | Abstract: Purpose: Hyperpolarized carbon-13 (13C) metabolic imaging is a noninvasive imaging modality for evaluating real-time metabolism. The purpose of this study was to develop and implement experimental strategies for using [1-13C]pyruvate to probe in vivo metabolism for patients with brain tumors and other neurological diseases. Methods: The 13C radiofrequency coils and pulse sequences were tested in a phantom and were performed using a 3 Tesla whole-body scanner. Samples of [1-13C]pyruvate were polarized using a SPINlab system. Dynamic 13C data were acquired from 8 patients previously diagnosed with brain tumors, who had received treatment and were being followed with serial magnetic resonance scans. Results: The phantom studies produced good-quality spectra with a reduction in signal intensity in the center attributed to the reception profiles of the 13C receive coils. Dynamic data obtained from a 3-cm slice through a patient's brain following injection with [1-13C]pyruvate showed the anticipated arrival of the agent, its conversion to lactate and bicarbonate, and subsequent reduction in signal intensity. A similar temporal pattern was observed in 2D dynamic patient studies, with signals corresponding to pyruvate, lactate, and bicarbonate being in normal appearing brain, but only pyruvate and lactate being detected in regions corresponding to the anatomical lesion. Physiological monitoring and follow-up confirmed that there were no adverse events associated with the injection. Conclusion: This study has presented the first application of hyperpolarized 13C metabolic imaging in patients with brain tumor and demonstrated the safety and feasibility of using hyperpolarized [1-13C]pyruvate to evaluate in vivo brain metabolism. Magn Reson Med 80:864–873, 2018. © 2018 International Society for Magnetic Resonance in Medicine.

Journal ArticleDOI
TL;DR: The purpose of this study was to accelerate the acquisition and reconstruction time of 3D magnetic resonance fingerprinting scans.
Abstract: Purpose The purpose of this study was to accelerate the acquisition and reconstruction time of 3D magnetic resonance fingerprinting scans. Methods A 3D magnetic resonance fingerprinting scan was accelerated by using a single-shot spiral trajectory with an undersampling factor of 48 in the x-y plane, and an interleaved sampling pattern with an undersampling factor of 3 through plane. Further acceleration came from reducing the waiting time between neighboring partitions. The reconstruction time was accelerated by applying singular value decomposition compression in k-space. Finally, a 3D premeasured B1 map was used to correct for the B1 inhomogeneity. Results The T1 and T2 values of the International Society for Magnetic Resonance in Medicine/National Institute of Standards and Technology MRI phantom showed a good agreement with the standard values, with an average concordance correlation coefficient of 0.99, and coefficient of variation of 7% in the repeatability scans. The results from in vivo scans also showed high image quality in both transverse and coronal views. Conclusions This study applied a fast acquisition scheme for a fully quantitative 3D magnetic resonance fingerprinting scan with a total acceleration factor of 144 as compared with the Nyquist rate, such that 3D T1 , T2 , and proton density maps can be acquired with whole-brain coverage at clinical resolution in less than 5 min. Magn Reson Med 79:2190-2197, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Journal ArticleDOI
TL;DR: To introduce a methodology for the reconstruction of multi‐shot, multi‐slice magnetic resonance imaging able to cope with both within‐plane and through‐plane rigid motion and to describe its application in structural brain imaging.
Abstract: Purpose To introduce a methodology for the reconstruction of multi-shot, multi-slice magnetic resonance imaging able to cope with both within-plane and through-plane rigid motion and to describe its application in structural brain imaging. Theory and Methods The method alternates between motion estimation and reconstruction using a common objective function for both. Estimates of three-dimensional motion states for each shot and slice are gradually refined by improving on the fit of current reconstructions to the partial k-space information from multiple coils. Overlapped slices and super-resolution allow recovery of through-plane motion and outlier rejection discards artifacted shots. The method is applied to T2 and T1 brain scans acquired in different views. Results The procedure has greatly diminished artifacts in a database of 1883 neonatal image volumes, as assessed by image quality metrics and visual inspection. Examples showing the ability to correct for motion and robustness against damaged shots are provided. Combination of motion corrected reconstructions for different views has shown further artifact suppression and resolution recovery. Conclusion The proposed method addresses the problem of rigid motion in multi-shot multi-slice anatomical brain scans. Tests on a large collection of potentially corrupted datasets have shown a remarkable image quality improvement. Magn Reson Med, 2017. © 2017 The Authors Magnetic Resonance in Medicine published by Wiley Periodicals, Inc. on behalf of International Society for Magnetic Resonance in Medicine. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Journal ArticleDOI
TL;DR: A 5D whole‐heart sparse imaging framework is proposed for simultaneous assessment of myocardial function and high‐resolution cardiac and respiratory motion‐resolved whole‐ heart anatomy in a single continuous noncontrast MR scan.
Abstract: Purpose A 5D whole-heart sparse imaging framework is proposed for simultaneous assessment of myocardial function and high-resolution cardiac and respiratory motion-resolved whole-heart anatomy in a single continuous noncontrast MR scan. Methods A non–electrocardiograph (ECG)-triggered 3D golden-angle radial balanced steady-state free precession sequence was used for data acquisition. The acquired 3D k-space data were sorted into a 5D dataset containing separated cardiac and respiratory dimensions using a self-extracted respiratory motion signal and a recorded ECG signal. Images were then reconstructed using XD-GRASP, a multidimensional compressed sensing technique exploiting correlations/sparsity along cardiac and respiratory dimensions. 5D whole-heart imaging was compared with respiratory motion-corrected 3D and 4D whole-heart imaging in nine volunteers for evaluation of the myocardium, great vessels, and coronary arteries. It was also compared with breath-held, ECG-gated 2D cardiac cine imaging for validation of cardiac function quantification. Results 5D whole-heart images received systematic higher quality scores in the myocardium, great vessels and coronary arteries. Quantitative coronary sharpness and length were always better for the 5D images. Good agreement was obtained for quantification of cardiac function compared with 2D cine imaging. Conclusion 5D whole-heart sparse imaging represents a robust and promising framework for simplified comprehensive cardiac MRI without the need for breath-hold and motion correction. Magn Reson Med 79:826–838, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Journal ArticleDOI
TL;DR: This paper, written by members of the Standards for Quantitative Magnetic Resonance committee, reviews standardization attempts and then details the need, requirements, and implementation plan for a standard system phantom for quantitative MRI.
Abstract: The MRI community is using quantitative mapping techniques to complement qualitative imaging. For quantitative imaging to reach its full potential, it is necessary to analyze measurements across systems and longitudinally. Clinical use of quantitative imaging can be facilitated through adoption and use of a standard system phantom, a calibration/standard reference object, to assess the performance of an MRI machine. The International Society of Magnetic Resonance in Medicine AdHoc Committee on Standards for Quantitative Magnetic Resonance was established in February 2007 to facilitate the expansion of MRI as a mainstream modality for multi-institutional measurements, including, among other things, multicenter trials. The goal of the Standards for Quantitative Magnetic Resonance committee was to provide a framework to ensure that quantitative measures derived from MR data are comparable over time, between subjects, between sites, and between vendors. This paper, written by members of the Standards for Quantitative Magnetic Resonance committee, reviews standardization attempts and then details the need, requirements, and implementation plan for a standard system phantom for quantitative MRI. In addition, application-specific phantoms and implementation of quantitative MRI are reviewed. Magn Reson Med 79:48-61, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Journal ArticleDOI
TL;DR: In this paper, a low-rank inverse problem was formulated for magnetic resonance fingerprinting and an alternating direction method of multipliers approach was proposed to reduce the number of Fourier transformations.
Abstract: The proposed reconstruction framework addresses the reconstruction accuracy, noise propagation and computation time for magnetic resonance fingerprinting. Based on a singular value decomposition of the signal evolution, magnetic resonance fingerprinting is formulated as a low rank (LR) inverse problem in which one image is reconstructed for each singular value under consideration. This LR approximation of the signal evolution reduces the computational burden by reducing the number of Fourier transformations. Also, the LR approximation improves the conditioning of the problem, which is further improved by extending the LR inverse problem to an augmented Lagrangian that is solved by the alternating direction method of multipliers. The root mean square error and the noise propagation are analyzed in simulations. For verification, in vivo examples are provided. The proposed LR alternating direction method of multipliers approach shows a reduced root mean square error compared to the original fingerprinting reconstruction, to a LR approximation alone and to an alternating direction method of multipliers approach without a LR approximation. Incorporating sensitivity encoding allows for further artifact reduction. The proposed reconstruction provides robust convergence, reduced computational burden and improved image quality compared to other magnetic resonance fingerprinting reconstruction approaches evaluated in this study. Magn Reson Med 79:83-96, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Journal ArticleDOI
TL;DR: In this feasibility study, a phase‐resolved functional lung imaging postprocessing method for extraction of dynamic perfusion and ventilation parameters using a conventional 1H lung MRI Fourier decomposition acquisition is introduced.
Abstract: Purpose In this feasibility study, a phase-resolved functional lung imaging postprocessing method for extraction of dynamic perfusion (Q) and ventilation (V) parameters using a conventional 1H lung MRI Fourier decomposition acquisition is introduced. Methods Time series of coronal gradient-echo MR images with a temporal resolution of 288 to 324 ms of two healthy volunteers, one patient with chronic thromboembolic hypertension, one patient with cystic fibrosis, and one patient with chronic obstructive pulmonary disease were acquired at 1.5 T. Using a sine model to estimate cardiac and respiratory phases of each image, all images were sorted to reconstruct full cardiac and respiratory cycles. Time to peak (TTP), V/Q maps, and fractional ventilation flow-volume loops were calculated. Results For the volunteers, homogenous ventilation and perfusion TTP maps (V-TTP, Q-TTP) were obtained. The chronic thromboembolic hypertension patient showed increased perfusion TTP in hypoperfused regions in visual agreement with dynamic contrast-enhanced MRI, which improved postpulmonary endaterectomy surgery. Cystic fibrosis and chronic obstructive pulmonary disease patients showed a pattern of increased V-TTP and Q-TTP in regions of hypoventilation and decreased perfusion. Fractional ventilation flow-volume loops of the chronic obstructive pulmonary disease patient were smaller in comparison with the healthy volunteer, and showed regional differences in visual agreement with functional small airways disease and emphysema on CT. Conclusions This study shows the feasibility of phase-resolved functional lung imaging to gain quantitative information regarding regional lung perfusion and ventilation without the need for ultrafast imaging, which will be advantageous for future clinical translation. Magn Reson Med 79:2306-2314, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Journal ArticleDOI
TL;DR: This work hypothesized that HRFv confounds FC estimates in the brain's default‐mode‐network and found that it did not.
Abstract: Purpose fMRI is the convolution of the hemodynamic response function (HRF) and unmeasured neural activity. HRF variability (HRFv) across the brain could, in principle, alter functional connectivity (FC) estimates from resting-state fMRI (rs-fMRI). Given that HRFv is driven by both neural and non-neural factors, it is problematic when it confounds FC. However, this aspect has remained largely unexplored even though FC studies have grown exponentially. We hypothesized that HRFv confounds FC estimates in the brain's default-mode-network. Methods We tested this hypothesis using both simulations (where the ground truth is known and modulated) as well as rs-fMRI data obtained in a 7T MRI scanner (N = 47, healthy). FC was obtained using 2 pipelines: data with hemodynamic deconvolution (DC) to estimate the HRF and minimize HRFv, and data with no deconvolution (NDC, HRFv-ignored). DC and NDC FC networks were compared, along with regional HRF differences, revealing potential false connectivities that resulted from HRFv. Results We found evidence supporting our hypothesis using both simulations and experimental data. With simulations, we found that HRFv could cause a change of up to 50% in FC. With rs-fMRI, several potential false connectivities attributable to HRFv, with majority connections being between different lobes, were identified. We found a double exponential relationship between the magnitude of HRFv and its impact on FC, with a mean/median error of 30.5/11.5% caused in FC by HRF confounds. Conclusion HRFv, if ignored, could cause identification of false FC. FC findings from HRFv-ignored data should be interpreted cautiously. We suggest deconvolution to minimize HRFv.

Journal ArticleDOI
TL;DR: To develop a fast magnetic resonance fingerprinting (MRF) method for quantitative chemical exchange saturation transfer (CEST) imaging.
Abstract: PURPOSE To develop a fast magnetic resonance fingerprinting (MRF) method for quantitative chemical exchange saturation transfer (CEST) imaging. METHODS We implemented a CEST-MRF method to quantify the chemical exchange rate and volume fraction of the Nα -amine protons of L-arginine (L-Arg) phantoms and the amide and semi-solid exchangeable protons of in vivo rat brain tissue. L-Arg phantoms were made with different concentrations (25-100 mM) and pH (pH 4-6). The MRF acquisition schedule varied the saturation power randomly for 30 iterations (phantom: 0-6 μT; in vivo: 0-4 μT) with a total acquisition time of ≤2 min. The signal trajectories were pattern-matched to a large dictionary of signal trajectories simulated using the Bloch-McConnell equations for different combinations of exchange rate, exchangeable proton volume fraction, and water T1 and T2 relaxation times. RESULTS The chemical exchange rates of the Nα -amine protons of L-Arg were significantly (P < 0.0001) correlated with the rates measured with the quantitation of exchange using saturation power method. Similarly, the L-Arg concentrations determined using MRF were significantly (P < 0.0001) correlated with the known concentrations. The pH dependence of the exchange rate was well fit (R2 = 0.9186) by a base catalyzed exchange model. The amide proton exchange rate measured in rat brain cortex (34.8 ± 11.7 Hz) was in good agreement with that measured previously with the water exchange spectroscopy method (28.6 ± 7.4 Hz). The semi-solid proton volume fraction was elevated in white (12.2 ± 1.7%) compared to gray (8.1 ± 1.1%) matter brain regions in agreement with previous magnetization transfer studies. CONCLUSION CEST-MRF provides a method for fast, quantitative CEST imaging.

Journal ArticleDOI
TL;DR: A quantitative technique to assess solute uptake into the brain parenchyma based on dynamic contrast‐enhanced MRI (DCE‐MRI) is proposed and whole brain gadolinium concentration maps are derived.
Abstract: Purpose We propose a quantitative technique to assess solute uptake into the brain parenchyma based on dynamic contrast-enhanced MRI (DCE-MRI). With this approach, a small molecular weight paramagnetic contrast agent (Gd-DOTA) is infused in the cerebral spinal fluid (CSF) and whole brain gadolinium concentration maps are derived. Methods We implemented a 3D variable flip angle spoiled gradient echo (VFA-SPGR) longitudinal relaxation time (T1) technique, the accuracy of which was cross-validated by way of inversion recovery rapid acquisition with relaxation enhancement (IR-RARE) using phantoms. Normal Wistar rats underwent Gd-DOTA infusion into CSF via the cisterna magna and continuous MRI for approximately 130 min using T1-weighted imaging. Dynamic Gd-DOTA concentration maps were calculated and parenchymal uptake was estimated. Results In the phantom study, T1 discrepancies between the VFA-SPGR and IR-RARE sequences were approximately 6% with a transmit coil inhomogeneity correction. In the in vivo study, contrast transport profiles indicated maximal parenchymal retention of approximately 19% relative to the total amount delivered into the cisterna magna. Conclusion Imaging strategies for accurate 3D contrast concentration mapping at 9.4T were developed and whole brain dynamic concentration maps were derived to study solute transport via the glymphatic system. The newly developed approach will enable future quantitative studies of the glymphatic system in health and disease states. Magn Reson Med, 2017. © 2017 International Society for Magnetic Resonance in Medicine.

Journal ArticleDOI
TL;DR: The purpose was to create a maximally time‐efficient and flexible diffusion acquisition capability with built‐in robustness to partially acquired or interrupted scans.
Abstract: Purpose: Advanced diffusion magnetic resonance imaging benefits from collecting as much data as is feasible but is highly sensitive to subject motion and the risk of data loss increases with longer acquisition times Our purpose was to create a maximally time-efficient and flexible diffusion acquisition capability with built-in robustness to partially acquired or interrupted scans Our framework has been developed for the developing Human Connectome Project, but different application domains are equally possible Methods: Complete flexibility in the sampling of diffusion space combined with free choice of phase-encode-direction and the temporal ordering of the sampling scheme was developed taking into account motion robustness, internal consistency, and hardware limits A split-diffusion-gradient preparation, multiband acceleration, and a restart capacity were added Results: The framework was used to explore different parameters choices for the desired high angular resolution diffusion imaging diffusion sampling For the developing Human Connectome Project, a high-angular resolution, maximally time-efficient (20 min) multishell protocol with 300 diffusion-weighted volumes was acquired in >400 neonates An optimal design of a high-resolution (12 × 12 mm2) two-shell acquisition with 54 diffusion weighted volumes was obtained using a split-gradient design Conclusion: The presented framework provides flexibility to generate time-efficient and motion-robust diffusion magnetic resonance imaging acquisitions taking into account hardware constraints that might otherwise result in sub-optimal choices Magn Reson Med, 2017 © 2017 The Authors Magnetic Resonance in Medicine published by Wiley Periodicals, Inc on behalf of International Society for Magnetic Resonance in Medicine This is an open access article under the terms of the Creative Commons Attribution NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes

Journal ArticleDOI
TL;DR: A method for converting Zero TE MR images into X‐ray attenuation information in the form of pseudo‐CT images is described and its performance for attenuation correction in PET/MR and dose planning in MR‐guided radiation therapy planning (RTP) is demonstrated.
Abstract: Purpose: To describe a method for converting Zero TE (ZTE) MR images into Xray attenuation information in the form of pseudo-CT images and demonstrate its performance for (1) attenuation correction ...

Journal ArticleDOI
TL;DR: CEST NMR or MRI experiments allow detection of low concentrated molecules with enhanced sensitivity via their proton exchange with the abundant water pool, but an exact quantification of the actual exchange rate is required to design optimal pulse sequences and/or specific sensitive agents.
Abstract: Purpose Chemical exchange saturation transfer (CEST) NMR or MRI experiments allow detection of low concentrated molecules with enhanced sensitivity via their proton exchange with the abundant water pool. Be it endogenous metabolites or exogenous contrast agents, an exact quantification of the actual exchange rate is required to design optimal pulse sequences and/or specific sensitive agents. Methods Refined analytical expressions allow deeper insight and improvement of accuracy for common quantification techniques. The accuracy of standard quantification methodologies, such as quantification of exchange rate using varying saturation power or varying saturation time, is improved especially for the case of nonequilibrium initial conditions and weak labeling conditions, meaning the saturation amplitude is smaller than the exchange rate (γB1 Results The improved analytical 'quantification of exchange rate using varying saturation power/time' (QUESP/QUEST) equations allow for more accurate exchange rate determination, and provide clear insights on the general principles to execute the experiments and to perform numerical evaluation. The proposed methodology was evaluated on the large-shift regime of paramagnetic chemical-exchange-saturation-transfer agents using simulated data and data of the paramagnetic Eu(III) complex of DOTA-tetraglycineamide. Conclusions The refined formulas yield improved exchange rate estimation. General convergence intervals of the methods that would apply for smaller shift agents are also discussed. Magn Reson Med 79:1708-1721, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Journal ArticleDOI
TL;DR: To develop parallel imaging techniques that simultaneously exploit coil sensitivity encoding, image phase prior information, similarities across multiple images, and complementary k‐space sampling for highly accelerated data acquisition.
Abstract: PURPOSE To develop parallel imaging techniques that simultaneously exploit coil sensitivity encoding, image phase prior information, similarities across multiple images, and complementary k-space sampling for highly accelerated data acquisition. METHODS We introduce joint virtual coil (JVC)-generalized autocalibrating partially parallel acquisitions (GRAPPA) to jointly reconstruct data acquired with different contrast preparations, and show its application in 2D, 3D, and simultaneous multi-slice (SMS) acquisitions. We extend the joint parallel imaging concept to exploit limited support and smooth phase constraints through Joint (J-) LORAKS formulation. J-LORAKS allows joint parallel imaging from limited autocalibration signal region, as well as permitting partial Fourier sampling and calibrationless reconstruction. RESULTS We demonstrate highly accelerated 2D balanced steady-state free precession with phase cycling, SMS multi-echo spin echo, 3D multi-echo magnetization-prepared rapid gradient echo, and multi-echo gradient recalled echo acquisitions in vivo. Compared to conventional GRAPPA, proposed joint acquisition/reconstruction techniques provide more than 2-fold reduction in reconstruction error. CONCLUSION JVC-GRAPPA takes advantage of additional spatial encoding from phase information and image similarity, and employs different sampling patterns across acquisitions. J-LORAKS achieves a more parsimonious low-rank representation of local k-space by considering multiple images as additional coils. Both approaches provide dramatic improvement in artifact and noise mitigation over conventional single-contrast parallel imaging reconstruction. Magn Reson Med 80:619-632, 2018. © 2018 International Society for Magnetic Resonance in Medicine.

Journal ArticleDOI
TL;DR: To determine the in vitro accuracy, test‐retest repeatability, and interplatform reproducibility of T1 quantification protocols used for dynamic contrast‐enhanced MRI at 1.5 and 3 T, six protocols were studied.
Abstract: Purpose To determine the in vitro accuracy, test-retest repeatability, and interplatform reproducibility of T1 quantification protocols used for dynamic contrast-enhanced MRI at 1.5 and 3 T. Methods A T1 phantom with 14 samples was imaged at eight centers with a common inversion-recovery spin-echo (IR-SE) protocol and a variable flip angle (VFA) protocol using seven flip angles, as well as site-specific protocols (VFA with different flip angles, variable repetition time, proton density, and Look-Locker inversion recovery). Factors influencing the accuracy (deviation from reference NMR T1 measurements) and repeatability were assessed using general linear mixed models. Interplatform reproducibility was assessed using coefficients of variation. Results For the common IR-SE protocol, accuracy (median error across platforms = 1.4–5.5%) was influenced predominantly by T1 sample (P < 10−6), whereas test-retest repeatability (median error = 0.2–8.3%) was influenced by the scanner (P < 10−6). For the common VFA protocol, accuracy (median error = 5.7–32.2%) was influenced by field strength (P = 0.006), whereas repeatability (median error = 0.7–25.8%) was influenced by the scanner (P < 0.0001). Interplatform reproducibility with the common VFA was lower at 3 T than 1.5 T (P = 0.004), and lower than that of the common IR-SE protocol (coefficient of variation 1.5T: VFA/IR-SE = 11.13%/8.21%, P = 0.028; 3 T: VFA/IR-SE = 22.87%/5.46%, P = 0.001). Among the site-specific protocols, Look-Locker inversion recovery and VFA (2–3 flip angles) protocols showed the best accuracy and repeatability (errors < 15%). Conclusions The VFA protocols with 2 to 3 flip angles optimized for different applications achieved acceptable balance of extensive spatial coverage, accuracy, and repeatability in T1 quantification (errors < 15%). Further optimization in terms of flip-angle choice for each tissue application, and the use of B1 correction, are needed to improve the robustness of VFA protocols for T1 mapping. Magn Reson Med, 2017. © 2017 International Society for Magnetic Resonance in Medicine.

Journal ArticleDOI
TL;DR: To develop a model‐based reconstruction technique for single‐shot T1 mapping with high spatial resolution, accuracy, and precision using an inversion‐recovery (IR) fast low‐angle shot (FLASH) acquisition with radial encoding.
Abstract: Purpose To develop a model-based reconstruction technique for single-shot T1 mapping with high spatial resolution, accuracy, and precision using an inversion-recovery (IR) fast low-angle shot (FLASH) acquisition with radial encoding. Methods The proposed model-based reconstruction jointly estimates all model parameters, that is, the equilibrium magnetization, steady-state magnetization, 1/ T1*, and all coil sensitivities from the data of a single-shot IR FLASH acquisition with a small golden-angle radial trajectory. Joint sparsity constraints on the parameter maps are exploited to improve the performance of the iteratively regularized Gauss-Newton method chosen for solving the nonlinear inverse problem. Validations include both a numerical and experimental T1 phantom, as well as in vivo studies of the human brain and liver at 3 T. Results In comparison to previous reconstruction methods for single-shot T1 mapping, which are based on real-time MRI with pixel-wise fitting and a model-based approach with a predetermination of coil sensitivities, the proposed method presents with improved robustness against phase errors and numerical precision in both phantom and in vivo studies. Conclusion The comprehensive model-based reconstruction with L1 regularization offers rapid and robust T1 mapping with high accuracy and precision. The method warrants accelerated computing and online implementation for extended clinical trials. Magn Reson Med 79:730–740, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Journal ArticleDOI
TL;DR: To build and evaluate a small‐footprint, lightweight, high‐performance 3T MRI scanner for advanced brain imaging with image quality that is equal to or better than conventional whole‐body clinical3T MRI scanners, while achieving substantial reductions in installation costs.
Abstract: Purpose To build and evaluate a small-footprint, lightweight, high-performance 3T MRI scanner for advanced brain imaging with image quality that is equal to or better than conventional whole-body clinical 3T MRI scanners, while achieving substantial reductions in installation costs. Methods A conduction-cooled magnet was developed that uses less than 12 liters of liquid helium in a gas-charged sealed system, and standard NbTi wire, and weighs approximately 2000 kg. A 42-cm inner-diameter gradient coil with asymmetric transverse axes was developed to provide patient access for head and extremity exams, while minimizing magnet-gradient interactions that adversely affect image quality. The gradient coil was designed to achieve simultaneous operation of 80-mT/m peak gradient amplitude at a slew rate of 700 T/m/s on each gradient axis using readily available 1-MVA gradient drivers. Results In a comparison of anatomical imaging in 16 patients using T2 -weighted 3D fluid-attenuated inversion recovery (FLAIR) between the compact 3T and whole-body 3T, image quality was assessed as equivalent to or better across several metrics. The ability to fully use a high slew rate of 700 T/m/s simultaneously with 80-mT/m maximum gradient amplitude resulted in improvements in image quality across EPI, DWI, and anatomical imaging of the brain. Conclusions The compact 3T MRI system has been in continuous operation at the Mayo Clinic since March 2016. To date, over 200 patient studies have been completed, including 96 comparison studies with a clinical 3T whole-body MRI. The increased gradient performance has reliably resulted in consistently improved image quality.

Journal ArticleDOI
TL;DR: An end‐to‐end deep convolutional neural network (CNN) based on deep residual network (ResNet) was proposed to efficiently reconstruct reliable T2 mapping from single‐shot overlapping‐echo detachment (OLED) planar imaging.
Abstract: Purpose An end-to-end deep convolutional neural network (CNN) based on deep residual network (ResNet) was proposed to efficiently reconstruct reliable T2 mapping from single-shot overlapping-echo detachment (OLED) planar imaging. Methods The training dataset was obtained from simulations that were carried out on SPROM (Simulation with PRoduct Operator Matrix) software developed by our group. The relationship between the original OLED image containing two echo signals and the corresponding T2 mapping was learned by ResNet training. After the ResNet was trained, it was applied to reconstruct the T2 mapping from simulation and in vivo human brain data. Results Although the ResNet was trained entirely on simulated data, the trained network was generalized well to real human brain data. The results from simulation and in vivo human brain experiments show that the proposed method significantly outperforms the echo-detachment-based method. Reliable T2 mapping with higher accuracy is achieved within 30 ms after the network has been trained, while the echo-detachment-based OLED reconstruction method took approximately 2 min. Conclusion The proposed method will facilitate real-time dynamic and quantitative MR imaging via OLED sequence, and deep convolutional neural network has the potential to reconstruct maps from complex MRI sequences efficiently.

Journal ArticleDOI
TL;DR: Microscopic diffusion anisotropy measurements from DDE promise greater specificity to changes in tissue microstructure compared with conventional diffusion tensor imaging, but implementation of DDE sequences on whole‐body MRI scanners is challenging because of the limited gradient strengths and lengthy acquisition times.
Abstract: Purpose The purpose of this study is to develop double diffusion encoding (DDE) MRI methods for clinical use. Microscopic diffusion anisotropy measurements from DDE promise greater specificity to changes in tissue microstructure compared with conventional diffusion tensor imaging, but implementation of DDE sequences on whole-body MRI scanners is challenging because of the limited gradient strengths and lengthy acquisition times. Methods A custom single-refocused DDE sequence was implemented on a 3T whole-body scanner. The DDE gradient orientation scheme and sequence parameters were optimized based on a Gaussian diffusion assumption. Using an optimized 5-min DDE acquisition, microscopic fractional anisotropy (μFA) maps were acquired for the first time in multiple sclerosis patients. Results Based on simulations and in vivo human measurements, six parallel and six orthogonal diffusion gradient pairs were found to be the minimum number of diffusion gradient pairs necessary to produce a rotationally invariant measurement of μFA. Simulations showed that optimal precision and accuracy of μFA measurements were obtained using b-values between 1500 and 3000 s/mm2 . The μFA maps showed improved delineation of multiple sclerosis lesions compared with conventional fractional anisotropy and distinct contrast from T2 -weighted fluid attenuated inversion recovery and T1 -weighted imaging. Conclusion The μFA maps can be measured using DDE in a clinical setting and may provide new opportunities for characterizing multiple sclerosis lesions and other types of tissue degeneration. Magn Reson Med 80:507-520, 2018. © 2017 International Society for Magnetic Resonance in Medicine.