Undersampling Scheme Research Articles

Accelerating the estimation of 3D spatially resolved T2 distributions

Obtaining quantitative, 3D spatially–resolved T2 distributions (T2 maps) from magnetic resonance data is of importance in both medical and porous media applications. Due to the long acquisition time, there is considerable interest in accelerating the experiments by applying undersampling schemes during the acquisition and developing reconstruction techniques for obtaining the 3D T2 maps from the undersampled data. A multi–echo spin echo pulse sequence is used in this work to acquire the undersampled data according to two different sampling patterns: a conventional coherent sampling pattern where the same set of lines in k-space is sampled for all equally-spaced echoes in the echo train, and a proposed incoherent sampling pattern where an independent set of k-space lines is sampled for each echo. The conventional reconstruction technique of total variation regularization is compared to the more recent techniques of nuclear norm regularization and Nuclear Total Generalized Variation (NTGV) regularization. It is shown that best reconstructions are obtained when the data acquired using an incoherent sampling scheme are processed using NTGV regularization. Using an incoherent sampling pattern and NTGV regularization as the reconstruction technique, quantitative results are obtained at sampling percentages as low as 3.1% of k–space, corresponding to a 32–fold decrease in the acquisition time, compared to a fully sampled dataset.

The Influence of Radial Undersampling Schemes on Compressed Sensing in Cardiac DTI.

Diffusion tensor imaging (DTI) is known to suffer from long acquisition time, which greatly limits its practical and clinical use. Undersampling of k-space data provides an effective way to reduce the amount of data to acquire while maintaining image quality. Radial undersampling is one of the most popular non-Cartesian k-space sampling schemes, since it has relatively lower sensitivity to motion than Cartesian trajectories, and artifacts from linear reconstruction are more noise-like. Therefore, radial imaging is a promising strategy of undersampling to accelerate acquisitions. The purpose of this study is to investigate various radial sampling schemes as well as reconstructions using compressed sensing (CS). In particular, we propose two randomly perturbed radial undersampling schemes: golden-angle and random angle. The proposed methods are compared with existing radial undersampling methods, including uniformity-angle, randomly perturbed uniformity-angle, golden-angle, and random angle. The results on both simulated and real human cardiac diffusion weighted (DW) images show that, for the same amount of k-space data, randomly sampling around a random radial line results in better reconstruction quality for DTI indices, such as fractional anisotropy (FA), mean diffusivities (MD), and that the randomly perturbed golden-angle undersampling yields the best results for cardiac CS-DTI image reconstruction.

Accelerated 4D phase contrast MRI in skeletal muscle contraction.

3D time-resolved (4D) phase contrast MRI can be used to study muscle contraction. However, 3D coverage with sufficient spatiotemporal resolution can only be achieved by interleaved acquisitions during many repetitions of the motion task, resulting in long scan times. The aim of this study was to develop a compressed sensing accelerated 4D phase contrast MRI technique for quantification of velocities and strain rate of the muscles in the lower leg during active plantarflexion/dorsiflexion. Nine healthy volunteers were scanned during active dorsiflexion/plantarflexion task. For each volunteer, we acquired a reference scan, as well as 4 different accelerated scans (k-space undersampling factors: 3.14X, 4.09X, 4.89X, and 6.41X) obtained using Cartesian Poisson disk undersampling schemes. The data was reconstructed using a compressed sensing pipeline. For each scan, velocity and strain rate values were quantified in the gastrocnemius lateralis, gastrocnemius medialis, tibialis anterior, and soleus. No significant differences in velocity values were observed as a function acceleration factor in the investigated muscles. The strain rate calculation resulted in one positive (s+ ) and one negative (s- ) eigenvalue, whereas the third eigenvalue (s3 ) was consistently 0 for all the acquisitions. No significant differences were observed for the strain rate eigenvalues as a function of acceleration factor. Data undersampling combined with compressed sensing reconstruction allowed obtainment of time-resolved phase contrast acquisitions with 3D coverage and quantitative information comparable to the reference scan. The 3D sensitivity of the method can help in understanding the connection between muscle architecture and muscle function in future studies.

A Noise-Filtered Under-Sampling Scheme for Imbalanced Classification.

Under-sampling is a popular data preprocessing method in dealing with class imbalance problems, with the purposes of balancing datasets to achieve a high classification rate and avoiding the bias toward majority class examples. It always uses full minority data in a training dataset. However, some noisy minority examples may reduce the performance of classifiers. In this paper, a new under-sampling scheme is proposed by incorporating a noise filter before executing resampling. In order to verify the efficiency, this scheme is implemented based on four popular under-sampling methods, i.e., Undersampling + Adaboost, RUSBoost, UnderBagging, and EasyEnsemble through benchmarks and significance analysis. Furthermore, this paper also summarizes the relationship between algorithm performance and imbalanced ratio. Experimental results indicate that the proposed scheme can improve the original undersampling-based methods with significance in terms of three popular metrics for imbalanced classification, i.e., the area under the curve, -measure, and -mean.

A Distance-Based Weighted Undersampling Scheme for Support Vector Machines and its Application to Imbalanced Classification.

A support vector machine (SVM) plays a prominent role in classic machine learning, especially classification and regression. Through its structural risk minimization, it has enjoyed a good reputation in effectively reducing overfitting, avoiding dimensional disaster, and not falling into local minima. Nevertheless, existing SVMs do not perform well when facing class imbalance and large-scale samples. Undersampling is a plausible alternative to solve imbalanced problems in some way, but suffers from soaring computational complexity and reduced accuracy because of its enormous iterations and random sampling process. To improve their classification performance in dealing with data imbalance problems, this work proposes a weighted undersampling (WU) scheme for SVM based on space geometry distance, and thus produces an improved algorithm named WU-SVM. In WU-SVM, majority samples are grouped into some subregions (SRs) and assigned different weights according to their Euclidean distance to the hyper plane. The samples in an SR with higher weight have more chance to be sampled and put to use in each learning iteration, so as to retain the data distribution information of original data sets as much as possible. Comprehensive experiments are performed to test WU-SVM via 21 binary-class and six multiclass publically available data sets. The results show that it well outperforms the state-of-the-art methods in terms of three popular metrics for imbalanced classification, i.e., area under the curve, F-Measure, and G-Mean.

3D diffusion-weighted 129 Xe MRI for whole lung morphometry.

PurposeTo obtain whole lung morphometry measurements from 129Xe in a single breath‐hold with 3D multiple b‐value 129Xe diffusion‐weighted MRI (DW‐MRI) with an empirically optimized diffusion time and compressed sensing for scan acceleration.MethodsProspective three‐fold undersampled 3D multiple b‐value hyperpolarized 129Xe DW‐MRI datasets were acquired, and the diffusion time (Δ) was iterated so as to provide diffusive length scale (LmD) estimates from the stretched exponential model (SEM) that are comparable to those from 3He. The empirically optimized 129Xe diffusion time was then implemented with a four‐fold undersampling scheme and was prospectively benchmarked against 3He measurements in a cohort of five healthy volunteers, six ex‐smokers, and two chronic obstructive pulmonary disease patients using both SEM‐derived LmD and cylinder model (CM)‐derived mean chord length (Lm).ResultsGood agreement between the mean 129Xe and 3He LmD (mean difference, 2.2%) and Lm (mean difference, 1.1%) values was obtained in all subjects at an empirically optimized 129Xe Δ = 8.5 ms.ConclusionCompressed sensing has facilitated single‐breath 3D multiple b‐value 129Xe DW‐MRI acquisitions, and results at 129Xe Δ = 8.5 ms indicate that 129Xe provides a viable alternative to 3He for whole lung morphometry mapping with either the SEM or CM. Magn Reson Med 79:2986–2995, 2018. © 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.

Performance of a fast and high-resolution multi-echo spin-echo sequence for prostate T2 mapping across multiple systems.

To evaluate the performance of a multi-echo spin-echo sequence with k-t undersampling scheme (k-t T2 ) in prostate cancer. Phantom experiments were performed at five systems to estimate the bias, short-term repeatability, and reproducibility across all systems expressed with the within-subject coefficient of variation (wCV). Monthly measurements were performed on two systems for long-term repeatability estimation. To evaluate clinical repeatability, two T2 maps (voxel size 0.8 × 0.8 × 3 mm3 ; 5 min) were acquired at separate visits on one system for 13 prostate cancer patients. Repeatability was assessed per patient in relation to spatial resolution. T2 values were compared for tumor, peripheral zone, and transition zone. Phantom measurements showed a small bias (median = -0.9 ms) and good short-term repeatability (median wCV = 0.5%). Long-term repeatability was 0.9 and 1.1% and reproducibility between systems was 1.7%. The median bias observed in patients was -1.1 ms. At voxel level, the median wCV was 15%, dropping to 4% for structures of 0.5 cm3 . The median tumor T2 values (79 ms) were significantly lower (P < 0.001) than in the peripheral zone (149 ms), but overlapped with the transition zone (91 ms). Reproducible T2 mapping of the prostate is feasible with good spatial resolution in a clinically reasonable scan time, allowing reliable measurement of T2 in structures as small as 0.5 cm3 . Magn Reson Med 79:1586-1594, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Quantitative DLA-based compressed sensing for T1-weighted acquisitions

High resolution Manganese Enhanced Magnetic Resonance Imaging (MEMRI), which uses manganese as a T1 contrast agent, has great potential for functional imaging of live neuronal tissue at single neuron scale. However, reaching high resolutions often requires long acquisition times which can lead to reduced image quality due to sample deterioration and hardware instability. Compressed Sensing (CS) techniques offer the opportunity to significantly reduce the imaging time. The purpose of this work is to test the feasibility of CS acquisitions based on Diffusion Limited Aggregation (DLA) sampling patterns for high resolution quantitative T1-weighted imaging. Fully encoded and DLA-CS T1-weighted images of Aplysia californica neural tissue were acquired on a 17.2T MRI system. The MR signal corresponding to single, identified neurons was quantified for both versions of the T1 weighted images. For a 50% undersampling, DLA-CS can accurately quantify signal intensities in T1-weighted acquisitions leading to only 1.37% differences when compared to the fully encoded data, with minimal impact on image spatial resolution. In addition, we compared the conventional polynomial undersampling scheme with the DLA and showed that, for the data at hand, the latter performs better. Depending on the image signal to noise ratio, higher undersampling ratios can be used to further reduce the acquisition time in MEMRI based functional studies of living tissues.

Compressed sensing‐based ground MTI with clutter rejection scheme for synthetic aperture radar

This study investigates moving target indication (MTI) and velocity estimation for the single-channel synthetic aperture radar system. A four-dimensional grid is employed for location and velocity representation of any moving targets. Clutter terms spread among all location grid cells and a small number of velocity grid cells corresponding to small velocities. Since in practical situations, clutter spectra have a certain widening in frequency, it is favourable to cancel an arbitrarily chosen number of clutter bins about zero Doppler. In this study, a subspace-based clutter rejection method is proposed to cancel the stationary ground clutter in addition to clutter up to higher Doppler frequencies. That is, after suppressing the clutter successfully, the ground MTI problem becomes sparse and a compressive sensing reconstruction algorithm is applied to solve for the location and velocity of the moving target. To reduce the amount of data storage and computational load, some undersampling schemes are done based on the ability of compressive sensing to solve the underdetermined set of equations. Finally, it is shown how the proposed clutter cancellation method is important for the high data rate reduction case of undersampling.

Highly-accelerated self-gated free-breathing 3D cardiac cine MRI: validation in assessment of left ventricular function.

This work presents a highly-accelerated, self-gated, free-breathing 3D cardiac cine MRI method for cardiac function assessment. A golden-ratio profile based variable-density, pseudo-random, Cartesian undersampling scheme was implemented for continuous 3D data acquisition. Respiratory self-gating was achieved by deriving motion signal from the acquired MRI data. A multi-coil compressed sensing technique was employed to reconstruct 4D images (3D+time). 3D cardiac cine imaging with self-gating was compared to bellows gating and the clinical standard breath-held 2D cine imaging for evaluation of self-gating accuracy, image quality, and cardiac function in eight volunteers. Reproducibility of 3D imaging was assessed. Self-gated 3D imaging provided an image quality score of 3.4±0.7 vs 4.0±0 with the 2D method (p=0.06). It determined left ventricular end-systolic volume as 42.4±11.5mL, end-diastolic volume as 111.1±24.7mL, and ejection fraction as 62.0±3.1%, which were comparable to the 2D method, with bias±1.96×SD of -0.8±7.5mL (p=0.90), 2.6±3.3mL (p=0.84) and 1.4±6.4% (p=0.45), respectively. The proposed 3D cardiac cine imaging method enables reliable respiratory self-gating performance with good reproducibility, and provides comparable image quality and functional measurements to 2D imaging, suggesting that self-gated, free-breathing 3D cardiac cine MRI framework is promising for improved patient comfort and cardiac MRI scan efficiency.

Sub-Nyquist sampling boosts targeted light transport through opaque scattering media.

Optical time-reversal techniques are being actively developed to focus light through or inside opaque scattering media. When applied to biological tissue, these techniques promise to revolutionize biophotonics by enabling deep-tissue non-invasive optical imaging, optogenetics, optical tweezing, and phototherapy. In all previous optical time-reversal experiments, the scattered light field was well-sampled during wavefront measurement and wavefront reconstruction, following the Nyquist sampling criterion. Here, we overturn this conventional practice by demonstrating that even when the scattered field is under-sampled, light can still be focused through or inside scattering media. Even more surprisingly, we show both theoretically and experimentally that the focus achieved by under-sampling can be one order of magnitude brighter than that achieved under the well-sampling conditions used in previous works, where 3×3 to 5×5 pixels were used to sample one speckle grain on average. Moreover, sub-Nyquist sampling improves the signal-to-noise ratio and the collection efficiency of the scattered light. We anticipate that this newly explored under-sampling scheme will transform the understanding of optical time reversal and boost the performance of optical imaging, manipulation, and communication through opaque scattering media.

First-pass myocardial perfusion MRI with reduced subendocardial dark-rim artifact using optimized Cartesian sampling.

The presence of subendocardial dark-rim artifact (DRA) remains an ongoing challenge in first-pass perfusion (FPP) cardiac magnetic resonance imaging (MRI). We propose a free-breathing FPP imaging scheme with Cartesian sampling that is optimized to minimize the DRA and readily enables near-instantaneous image reconstruction. The proposed FPP method suppresses Gibbs ringing effects-a major underlying factor for the DRA-by "shaping" the underlying point spread function through a two-step process: 1) an undersampled Cartesian sampling scheme that widens the k-space coverage compared to the conventional scheme; and 2) a modified parallel-imaging scheme that incorporates optimized apodization (k-space data filtering) to suppress Gibbs-ringing effects. Healthy volunteer studies (n = 10) were performed to compare the proposed method against the conventional Cartesian technique-both using a saturation-recovery gradient-echo sequence at 3T. Furthermore, FPP imaging studies using the proposed method were performed in infarcted canines (n = 3), and in two symptomatic patients with suspected coronary microvascular dysfunction for assessment of myocardial hypoperfusion. Width of the DRA and the number of DRA-affected myocardial segments were significantly reduced in the proposed method compared to the conventional approach (width: 1.3 vs. 2.9 mm, P < 0.001; number of segments: 2.6 vs. 8.7; P < 0.0001). The number of slices with severe DRA was markedly lower for the proposed method (by 10-fold). The reader-assigned image quality scores were similar (P = 0.2), although the quantified myocardial signal-to-noise ratio was lower for the proposed method (P < 0.05). Animal studies showed that the proposed method can detect subendocardial perfusion defects and patient results were consistent with the gold-standard invasive test. The proposed free-breathing Cartesian FPP imaging method significantly reduces the prevalence of severe DRAs compared to the conventional approach while maintaining similar resolution and image quality. 2 J. Magn. Reson. Imaging 2017;45:542-555.

WE‐AB‐207A‐04: Random Undersampled Cone Beam CT: Theoretical Analysis and a Novel Reconstruction Method

Purpose:Reducing x‐ray exposure and speeding up data acquisition motived studies on projection data undersampling. It is an important question that for a given undersampling ratio, what the optimal undersampling approach is. In this study, we propose a new undersampling scheme: random‐ray undersampling. We will mathematically analyze its projection matrix properties and demonstrate its advantages. We will also propose a new reconstruction method that simultaneously performs CT image reconstruction and projection domain data restoration.Methods:By representing projection operator under the basis of singular vectors of full projection operator, matrix representations for an undersampling case can be generated and numerical singular value decomposition can be performed. We compared properties of matrices among three undersampling approaches: regular‐view undersampling, regular‐ray undersampling, and the proposed random‐ray undersampling. To accomplish CT reconstruction for random undersampling, we developed a novel method that iteratively performs CT reconstruction and missing projection data restoration via regularization approaches.Results:For a given undersampling ratio, random‐ray undersampling preserved mathematical properties of full projection operator better than the other two approaches. This translates to advantages of reconstructing CT images at lower errors. Different types of image artifacts were observed depending on undersampling strategies, which were ascribed to the unique singular vectors of the sampling operators in the image domain. We tested the proposed reconstruction algorithm on a Forbid phantom with only 30% of the projection data randomly acquired. Reconstructed image error was reduced from 9.4% in a TV method to 7.6% in the proposed method.Conclusion:The proposed random‐ray undersampling is mathematically advantageous over other typical undersampling approaches. It may permit better image reconstruction at the same undersampling ratio. The novel algorithm suitable for this random‐ray undersampling was able to reconstruct high‐quality images.

Differential wall shear stress patterns in cerebral circulation of prediabetic adults: Possible evidence for increased stroke risk?

Proper endothelial function is crucial to cerebral vascular health and is affected by wall shear stress (WSS) patterns. Metabolic syndrome (MetSyn) is a prediabetic state that contributes to endothelial dysfunction which can lead to clinical conditions, such as, stroke. The aim of this study was to quantify and compare resting WSS in multiple conduit cerebral arteries in healthy and MetSyn subjects. We hypothesized adults with Metsyn will display reduced WSS in cerebral arteries when compared to healthy controls. 7 healthy controls (29±2y) and 7 MetSyn (37±3y) underwent a single four‐dimensional (4D) flow MRI scan during quiet rest in the supine position. Data acquisition was based on a radial undersampling scheme named Phase Contrast Vastly‐undersampled Isotropic Projection Reconstruction (PC VIPR) used to generate a phase contrast (PC) angiogram, as well as, measure cerebral blood flow (CBF), and WSS non‐invasively without use of a contrast agent (three directional velocity encoding at a spatial and temporal resolution of 0.68 mm and 20 cardiac phases, respectively). Average scan time was ~ 5 minutes and produced large imaging volumes. Nine arteries were analyzed: Middle cerebral (MCA) ×2, anterior cerebral (ACA) ×2, internal carotid (ICA) ×2, vertebral (VA) ×2, and basilar. WSS was measured by segmenting the vascular tree from the PC angiograms and subsequently calculating the velocity gradient at the vessel wall from the 4D Flow MRI data over ~ 3 voxels. Total CBF was calculated as the sum of the left and right ICA and basilar artery. Resting total CBF was 577±45 ml· min−1 in healthy controls and 499±46 ml ·min−1 in MetSyn (p=0.25). Metsyn exhibited lower WSS (Pa) in the left MCA (3.7±0.2 vs. 3.0±0.2; p=0.01), right ACA (3.4±0.2 vs. 2.9±0.2; p=0.05), right ICA (2.7±0.3 vs. 1.8±0.1; p=0.01) and left ICA (2.8±0.3 vs. 1.5±0.1; p&lt;0.01). This non‐invasive MRI approach provides new insight into local flow and shear patterns that might have a large impact on cerebral endothelial function. Younger adults with MetSyn exhibit lower WSS in anterior portions of the brain. Furthermore, these findings are consistent with the observation that stroke occurs most commonly in the anterior cerebral circulation. Given the young age (&lt; 45 y) of these subjects, we propose chronically reduced WSS leads to endothelial dysfunction in middle/old age, thereby contributing to increased risk and severity of stroke in MetSyn.Support or Funding InformationFunded by the American Diabetes Association and an American Physiological Society Undergraduate Research Summer Fellowship.

4D Contrast-enhanced MR Angiography with the Keyhole Technique in Children: Technique and Clinical Applications.

Unlike in adults, contrast agent-enhanced magnetic resonance (MR) angiography in the pediatric population raises unique challenges such as faster heart rates, more rapid arteriovenous transit, smaller structures, smaller volumes of contrast agent used, and more complex disease processes. A need exists for a rapid contrast-enhanced MR angiographic technique that can separate the arterial and venous phases of contrast enhancement in sedated pediatric patients breathing freely during the course of an examination. In time-resolved contrast-enhanced MR angiography with the keyhole method (four-dimensional [4D] contrast-enhanced MR angiography), various spatial and temporal frequency undersampling schemes are used to substantially reduce the time of acquisition without markedly compromising spatial resolution. The keyhole method can be briefly described as an undersampling approach in which only a small region of the k-space (keyhole) around the center is repeatedly sampled while the periphery is sampled only once during acquisition. This method provides a wide range of options that can be used to overcome conventional limitations of contrast-enhanced MR angiography in children and opens the door for several new pediatric applications, including evaluation of congenital heart disease in neonates and infants, thoracic and extremity vascular pathologic conditions, high-flow vascular malformations, systemic vein thrombosis, and pediatric portal hypertension. This review provides a technical overview of 4D contrast-enhanced MR angiography, outlines its advantages and pitfalls in the pediatric population, and also describes various applications in children, including modifications of the technique needed for each application.

Influence of temporal regularization and radial undersampling factor on compressed sensing reconstruction in dynamic contrast enhanced MRI of the breast.

To evaluate the influence of temporal sparsity regularization and radial undersampling on compressed sensing reconstruction of dynamic contrast-enhanced (DCE) MRI, using the iterative Golden-angle RAdial Sparse Parallel (iGRASP) MRI technique in the setting of breast cancer evaluation. DCE-MRI examinations of the breast (n = 7) were conducted using iGRASP at 3 Tesla. Images were reconstructed with five different radial undersampling schemes corresponding to temporal resolutions between 2 and 13.4 s/frame and with four different weights for temporal sparsity regularization (λ = 0.1, 0.5, 2, and 6 times of noise level). Image similarity to time-averaged reference images was assessed by two breast radiologists and using quantitative metrics. Temporal similarity was measured in terms of wash-in slope and contrast kinetic model parameters. iGRASP images reconstructed with λ = 2 and 5.1 s/frame had significantly (P < 0.05) higher similarity to time-averaged reference images than the images with other reconstruction parameters (mutual information (MI) >5%), in agreement with the assessment of two breast radiologists. Higher undersampling (temporal resolution < 5.1 s/frame) required stronger temporal sparsity regularization (λ ≥ 2) to remove streaking aliasing artifacts (MI > 23% between λ = 2 and 0.5). The difference between the kinetic-model transfer rates of benign and malignant groups decreased as temporal resolution decreased (82% between 2 and 13.4 s/frame). This study demonstrates objective spatial and temporal similarity measures can be used to assess the influence of sparsity constraint and undersampling in compressed sensing DCE-MRI and also shows that the iGRASP method provides the flexibility of optimizing these reconstruction parameters in the postprocessing stage using the same acquired data.

Volumetric spiral chemical shift imaging of hyperpolarized [2-(13) c]pyruvate in a rat c6 glioma model.

MRS of hyperpolarized [2-(13)C]pyruvate can be used to assess multiple metabolic pathways within mitochondria as the (13)C label is not lost with the conversion of pyruvate to acetyl-CoA. This study presents the first MR spectroscopic imaging of hyperpolarized [2-(13)C]pyruvate in glioma-bearing brain. Spiral chemical shift imaging with spectrally undersampling scheme (1042 Hz) and a hard-pulse excitation was exploited to simultaneously image [2-(13)C]pyruvate, [2-(13)C]lactate, and [5-(13)C]glutamate, the metabolites known to be produced in brain after an injection of hyperpolarized [2-(13)C]pyruvate, without chemical shift displacement artifacts. A separate undersampling scheme (890 Hz) was also used to image [1-(13)C]acetyl-carnitine. Healthy and C6 glioma-implanted rat brains were imaged at baseline and after dichloroacetate administration, a drug that modulates pyruvate dehydrogenase kinase activity. The baseline metabolite maps showed higher lactate and lower glutamate in tumor as compared to normal-appearing brain. Dichloroacetate led to an increase in glutamate in both tumor and normal-appearing brain. Dichloroacetate-induced %-decrease of lactate/glutamate was comparable to the lactate/bicarbonate decrease from hyperpolarized [1-(13)C]pyruvate studies. Acetyl-carnitine was observed in the muscle/fat tissue surrounding the brain. Robust volumetric imaging with hyperpolarized [2-(13)C]pyruvate and downstream products was performed in glioma-bearing rat brains, demonstrating changes in mitochondrial metabolism with dichloroacetate.

Rapid acquisition of helium-3 and proton three-dimensional image sets of the human lung in a single breath-hold using compressed sensing.

To develop and validate a method for acquiring helium-3 ((3) He) and proton ((1) H) three-dimensional (3D) image sets of the human lung with isotropic spatial resolution within a 10-s breath-hold by using compressed sensing (CS) acceleration, and to assess the fidelity of undersampled images compared with fully sampled images. The undersampling scheme for CS acceleration was optimized and tested using (3) He ventilation data. Rapid 3D acquisition of both (3) He and (1) H data during one breath-hold was then implemented, based on a balanced steady-state free-precession pulse sequence, by random undersampling of k-space with reconstruction by means of minimizing the L1 norm and total variance. CS-reconstruction fidelity was evaluated quantitatively by comparing fully sampled and retrospectively undersampled image sets. Helium-3 and (1) H 3D image sets of the lung with isotropic 3.9-mm resolution were acquired during a single breath-hold in 12 s and 8 s using acceleration factors of 2 and 3, respectively. Comparison of fully sampled and retrospectively undersampled (3) He and (1) H images yielded mean absolute errors <10% and structural similarity indices >0.9. By randomly undersampling k-space and using CS reconstruction, high-quality (3) He and (1) H 3D image sets with isotropic 3.9-mm resolution can be acquired within an 8-s breath-hold.

Bernoulli-based random undersampling schemes for 2D seismic data regularization

Seismic data regularization is an important preprocessing step in seismic signal processing. Traditional seismic acquisition methods follow the Shannon-Nyquist sampling theorem, whereas compressive sensing (CS) provides a fundamentally new paradigm to overcome limitations in data acquisition. Besides the sparse representation of seismic signal in some transform domain and the 1-norm reconstruction algorithm, the seismic data regularization quality of CS-based techniques strongly depends on random undersampling schemes. For 2D seismic data, discrete uniform-based methods have been investigated, where some seismic traces are randomly sampled with an equal probability. However, in theory and practice, some seismic traces with different probability are required to be sampled for satisfying the assumptions in CS. Therefore, designing new undersampling schemes is imperative. We propose a Bernoulli-based random undersampling scheme and its jittered version to determine the regular traces that are randomly sampled with different probability, while both schemes comply with the Bernoulli process distribution. We performed experiments using the Fourier and curvelet transforms and the spectral projected gradient reconstruction algorithm for 1-norm (SPGL1), and ten different random seeds. According to the signal-to-noise ratio (SNR) between the original and reconstructed seismic data, the detailed experimental results from 2D numerical and physical simulation data show that the proposed novel schemes perform overall better than the discrete uniform schemes.

Highly accelerated aortic 4D flow MR imaging with variable-density random undersampling

PurposeTo investigate an effective time-resolved variable-density random undersampling scheme combined with an efficient parallel image reconstruction method for highly accelerated aortic 4D flow MR imaging with high reconstruction accuracy. Materials and MethodsVariable-density Poisson-disk sampling (vPDS) was applied in both the phase-slice encoding plane and the temporal domain to accelerate the time-resolved 3D Cartesian acquisition of flow imaging. In order to generate an improved initial solution for the iterative self-consistent parallel imaging method (SPIRiT), a sample-selective view sharing reconstruction for time-resolved random undersampling (STIRRUP) was introduced. The performance of different undersampling and image reconstruction schemes were evaluated by retrospectively applying those to fully sampled data sets obtained from three healthy subjects and a flow phantom. ResultsUndersampling pattern based on the combination of time-resolved vPDS, the temporal sharing scheme STIRRUP, and parallel imaging SPIRiT, were able to achieve 6-fold accelerated 4D flow MRI with high accuracy using a small number of coils (N=5). The normalized root mean square error between aorta flow waveforms obtained with the acceleration method and the fully sampled data in three healthy subjects was 0.04±0.02, and the difference in peak-systolic mean velocity was −0.29±2.56cm/s. ConclusionQualitative and quantitative evaluation of our preliminary results demonstrate that time-resolved variable-density random sampling is efficient for highly accelerating 4D flow imaging while maintaining image reconstruction accuracy.