John M. Pauly

Sparse MRI: The Application of Compressed Sensing for Rapid MR Imaging

The sparsity which is implicit in MR images is exploited to significantly undersample k‐space. Some MR images such as angiograms are already sparse in the pixel representation; other, more complicated images have a sparse representation in some transform domain–for example, in terms of spatial finite‐differences or their wavelet coefficients. According to the recently developed mathematical theory of compressed‐sensing, images with a sparse representation can be recovered from randomly undersampled k‐space data, provided an appropriate nonlinear recovery scheme is used. Intuitively, artifacts due to random undersampling add as noise‐like interference. In the sparse transform domain the significant coefficients stand out above the interference. A nonlinear thresholding scheme can recover the sparse coefficients, effectively recovering the image itself. In this article, practical incoherent undersampling schemes are developed and analyzed by means of their aliasing interference. Incoherence is introduced by pseudo‐random variable‐density undersampling of phase‐encodes. The reconstruction is performed by minimizing the ℓ1 norm of a transformed image, subject to data fidelity constraints. Examples demonstrate improved spatial resolution and accelerated acquisition for multislice fast spin‐echo brain imaging and 3D contrast enhanced angiography. Magn Reson Med, 2007.

Compressed Sensing MRI

This article reviews the requirements for successful compressed sensing (CS), describes their natural fit to MRI, and gives examples of four interesting applications of CS in MRI. The authors emphasize on an intuitive understanding of CS by describing the CS reconstruction as a process of interference cancellation. There is also an emphasis on the understanding of the driving factors in applications, including limitations imposed by MRI hardware, by the characteristics of different types of images, and by clinical concerns.

Parameter relations for the Shinnar-Le Roux selective excitation pulse design algorithm (NMR imaging)

An overview of the Shinnar-Le Roux (SLR) algorithm is presented. It is shown how the performance of SLR pulses can be very accurately specified analytically. This reveals how to design a pulse that produces a specified slice profile and allows the pulse designer to trade off analytically the parameters describing the pulse performance. Several examples are presented to illustrate the more important tradeoffs. These include linear-phase and minimum- and maximum-phase pulses. Linear-phase pulses can be refocused with a gradient reversal and can be used as spin-echo pulses. Minimum- and maximum-phase pulses have better slice profiles, but cannot be completely refocused.

A k-space analysis of small-tip-angle excitation

Abstract We present here a method for analyzing selective excitation in terms of spatial frequency ( k ) space. Using this analysis we show how to design inherently refocused selective excitation pulses in one and two dimensions. The analysis is based on a small-tip model, but holds well for 90° tip angles.

Positive contrast magnetic resonance imaging of cells labeled with magnetic nanoparticles.

Contrast agents incorporating superparamagnetic iron‐oxide nanoparticles have shown promise as a means to visualize labeled cells using MRI. Labeled cells cause significant signal dephasing due to the magnetic field inhomogeneity induced in water molecules near the cell. With the resulting signal void as the means for detection, the particles behave as a negative contrast agent, which can suffer from partial‐volume effects. In this paper, a new method is described for imaging labeled cells with positive contrast. Spectrally selective RF pulses are used to excite and refocus the off‐resonance water surrounding the labeled cells so that only the fluid and tissue immediately adjacent to the labeled cells are visible in the image. Phantom, in vitro, and in vivo experiments show the feasibility of the new method. A significant linear correlation (r = 0.87, P < 0.005) between the estimated number of cells and the signal was observed. Magn Reson Med 53:999–1005, 2005.

SPIRiT: Iterative self-consistent parallel imaging reconstruction from arbitrary k-space

A new approach to autocalibrating, coil‐by‐coil parallel imaging reconstruction, is presented. It is a generalized reconstruction framework based on self‐consistency. The reconstruction problem is formulated as an optimization that yields the most consistent solution with the calibration and acquisition data. The approach is general and can accurately reconstruct images from arbitrary k‐space sampling patterns. The formulation can flexibly incorporate additional image priors such as off‐resonance correction and regularization terms that appear in compressed sensing. Several iterative strategies to solve the posed reconstruction problem in both image and k‐space domain are presented. These are based on a projection over convex sets and conjugate gradient algorithms. Phantom and in vivo studies demonstrate efficient reconstructions from undersampled Cartesian and spiral trajectories. Reconstructions that include off‐resonance correction and nonlinear ℓ1‐wavelet regularization are also demonstrated. Magn Reson Med, 2010.

Learned regulation of spatially localized brain activation using real-time fMRI.

It is not currently known whether subjects can learn to voluntarily control activation in localized regions of their own brain using neuroimaging. Here, we show that subjects were able to learn enhanced voluntary control over task-specific activation in a chosen target region, the somatomotor cortex. During an imagined manual action task, subjects were provided with continuous direction regarding their cognitive processes. Subjects received feedback information about their current level of activation in a target region of interest (ROI) derived using real-time functional magnetic resonance imaging (rtfMRI), and they received automatically-adjusted instructions for the level of activation to achieve. Information was provided both as continously upated graphs and using a simple virtual reality interface that provided an image analog of the level of activation. Through training, subjects achieved an enhancement in their control over brain activation that was anatomically specific to the target ROI, the somatomotor cortex. The enhancement took place when rtfMRI-based training was provided, but not in a control group that received similar training without rtfMRI information, showing that the effect was not due to conventional, practice-based neural plasticity alone. Following training, using cognitive processes alone subjects could volitionally induce fMRI activation in the somatomotor cortex that was comparable in magnitude to the activation observed during actual movement. The trained subjects increased fMRI activation without muscle tensing, and were able to continue to control brain activation even when real-time fMRI information was no longer provided. These results show that rtfMRI information can be used to direct cognitive processes, and that subjects are able to learn volitionally regulate activation in an anatomically-targeted brain region, surpassing the task-driven activation present before training.

A homogeneity correction method for magnetic resonance imaging with time-varying gradients

When time-varying gradients are used for imaging, the off-resonance behavior does not just cause geometric distortion as is the case with spin-warp imaging, but changes the shape of the impulse response and causes blurring. This effect is well known for projection reconstruction and spiral k-space scanning sequences. The authors introduce a reconstruction and homogeneity correction method to correct for the zeroth order effects of inhomogeneity using prior knowledge of the inhomogeneity. In this method, the data are segmented according to collection time, reconstructed using some fast, linear algorithm, correlated for inhomogeneity, and then superimposed to yield a homogeneity corrected image. This segmented method is compared to a conjugate phase reconstruction in terms of degree of correction and execution time. The authors apply this method to in vivo images using projection-reconstruction and spiral-scan sequences.

Saturated double-angle method for rapid B1+ mapping

For in vivo magnetic resonance imaging at high field (≥3 T) it is essential to consider the homogeneity of the active B1 field (B1+), particularly if surface coils are used for RF transmission. A new method is presented for highly rapid B1+ magnitude mapping. It combines the double angle method with a B1‐insensitive magnetization‐reset sequence such that the choice of repetition time (TR) is independent of T1 and with a multislice segmented (spiral) acquisition to achieve volumetric coverage with adequate spatial resolution in a few seconds. Phantom experiments confirmed the accuracy of this technique even when TR ≪ T1, with the side effect being lowered SNR. The speed of this method enabled B1+ mapping in the chest and abdomen within a single breath‐hold. In human cardiac imaging, the method enabled whole‐heart coverage within a single 16‐s breath‐hold. Results from phantoms and healthy volunteers at 1.5 T and 3 T are presented. Magn Reson Med, 2006.

SEMAC: Slice Encoding for Metal Artifact Correction in MRI

Magnetic resonance imaging (MRI) near metallic implants remains an unmet need because of severe artifacts, which mainly stem from large metal‐induced field inhomogeneities. This work addresses MRI near metallic implants with an innovative imaging technique called “Slice Encoding for Metal Artifact Correction” (SEMAC). The SEMAC technique corrects metal artifacts via robust encoding of each excited slice against metal‐induced field inhomogeneities. The robust slice encoding is achieved by extending a view‐angle‐tilting (VAT) spin‐echo sequence with additional z‐phase encoding. Although the VAT compensation gradient suppresses most in‐plane distortions, the z‐phase encoding fully resolves distorted excitation profiles that cause through‐plane distortions. By positioning all spins in a region‐of‐interest to their actual spatial locations, the through‐plane distortions can be corrected by summing up the resolved spins in each voxel. The SEMAC technique does not require additional hardware and can be deployed to the large installed base of whole‐body MRI systems. The efficacy of the SEMAC technique in eliminating metal‐induced distortions with feasible scan times is validated in phantom and in vivo spine and knee studies. Magn Reson Med 62:66–76, 2009.