Autocalibrated coil sensitivity estimation for parallel imaging

Accelerated MR spectroscopic imaging-a review of current and emerging techniques

Over more than 30 years in vivo MR spectroscopic imaging (MRSI) has undergone an enormous evolution from theoretical concepts in the early 1980s to the robust imaging technique that it is today. The development of both fast and efficient sampling and reconstruction techniques has played a fundamental role in this process. State-of-the-art MRSI has grown from a slow purely phase-encoded acquisition technique to a method that today combines the benefits of different acceleration techniques. These include shortening of repetition times, spatial-spectral encoding, undersampling of k-space and time domain, and use of spatial-spectral prior knowledge in the reconstruction. In this way in vivo MRSI has considerably advanced in terms of spatial coverage, spatial resolution, acquisition speed, artifact suppression, number of detectable metabolites and quantification precision. Acceleration not only has been the enabling factor in high-resolution whole-brain 1 H-MRSI, but today is also common in non-proton MRSI (31 P, 2 H and 13 C) and applied in many different organs. In this process, MRSI techniques had to constantly adapt, but have also benefitted from the significant increase of magnetic field strength boosting the signal-to-noise ratio along with high gradient fidelity and high-density receive arrays. In combination with recent trends in image reconstruction and much improved computation power, these advances led to a number of novel developments with respect to MRSI acceleration. Today MRSI allows for non-invasive and non-ionizing mapping of the spatial distribution of various metabolites' tissue concentrations in animals or humans, is applied for clinical diagnostics and has been established as an important tool for neuro-scientific and metabolism research. This review highlights the developments of the last five years and puts them into the context of earlier MRSI acceleration techniques. In addition to 1 H-MRSI it also includes other relevant nuclei and is not limited to certain body regions or specific applications.

Ultra-high resolution blood volume fMRI and BOLD fMRI in humans at 9.4 T: Capabilities and challenges

Functional mapping of cerebral blood volume (CBV) changes has the potential to reveal brain activity with high localization specificity at the level of cortical layers and columns. Non-invasive CBV imaging using Vascular Space Occupancy (VASO) at ultra-high magnetic field strengths promises high spatial specificity but poses unique challenges in human applications. As such, 9.4 T B1+ and B0 inhomogeneities limit efficient blood tagging, while the specific absorption rate (SAR) constraints limit the application of VASO-specific RF pulses. Moreover, short T2* values at 9.4 T require short readout duration, and long T1 values at 9.4 T can cause blood-inflow contaminations. In this study, we investigated the applicability of layer-dependent CBV-fMRI at 9.4 T in humans. We addressed the aforementioned challenges by combining multiple technical advancements: temporally alternating pTx B1+ shimming parameters, advanced adiabatic RF-pulses, 3D-EPI signal readout, optimized GRAPPA acquisition and reconstruction, and stability-optimized RF channel combination. We found that a combination of suitable advanced methodology alleviates the challenges and potential artifacts, and that VASO fMRI provides reliable measures of CBV change across cortical layers in humans at 9.4 T. The localization specificity of CBV-fMRI, combined with the high sensitivity of 9.4 T, makes this method an important tool for future studies investigating cortical micro-circuitry in humans.

Dynamic 2D self-phase-map Nyquist ghost correction for simultaneous multi-slice echo planar imaging

PURPOSE

To develop a reconstruction pipeline that intrinsically accounts for both simultaneous multislice echo planar imaging (SMS-EPI) reconstruction and dynamic slice-specific Nyquist ghosting correction in time-series data.

METHODS

After 1D slice-group average phase correction, the separate polarity (i.e., even and odd echoes) SMS-EPI data were unaliased by slice GeneRalized Autocalibrating Partial Parallel Acquisition. Both the slice-unaliased even and odd echoes were jointly reconstructed using a model-based framework, extended for SMS-EPI reconstruction that estimates a 2D self-phase map, corrects dynamic slice-specific phase errors, and combines data from all coils and echoes to obtain the final images.

RESULTS

The percentage ghost-to-signal ratios (%GSRs) and its temporal variations for MB3R_y 2 with a field of view/4 shift in a human brain obtained by the proposed dynamic 2D and standard 1D phase corrections were 1.37 ± 0.11 and 2.66 ± 0.16, respectively. Even with a large regularization parameter λ applied in the proposed reconstruction, the smoothing effect in fMRI activation maps was comparable to a very small Gaussian kernel size 1 × 1 × 1 mm³ .

CONCLUSION

The proposed reconstruction pipeline reduced slice-specific phase errors in SMS-EPI, resulting in reduction of GSR. It is applicable for functional MRI studies because the smoothing effect caused by the regularization parameter selection can be minimal in a blood-oxygen-level-dependent activation map.

Techniques for blood volume fMRI with VASO: From low-resolution mapping towards sub-millimeter layer-dependent applications

Quantitative cerebral blood volume (CBV) fMRI has the potential to overcome several specific limitations of BOLD fMRI. It provides direct physiological interpretability and promises superior localization specificity in applications of sub-millimeter resolution fMRI applications at ultra-high magnetic fields (7T and higher). Non-invasive CBV fMRI using VASO (vascular space occupancy), however, is inherently limited with respect to its data acquisition efficiency, restricting its imaging coverage and achievable spatial and temporal resolution. This limitation may be reduced with recent advanced acceleration and reconstruction strategies that allow two-dimensional acceleration, such as in simultaneous multi-slice (SMS) 2D-EPI or 3D-EPI in combination with CAIPIRINHA field-of-view shifting. In this study, we sought to determine the functional sensitivity and specificity of these readout strategies with VASO over a broad range of spatial resolutions; spanning from low spatial resolution (3mm) whole-cortex to sub-millimeter (0.75mm) slab-of-cortex (for cortical layer-dependent applications). In the thermal-noise-dominated regime of sub-millimeter resolutions, 3D-EPI-VASO provides higher temporal stability and sensitivity to detect changes in CBV compared to 2D-EPI-VASO. In this regime, 3D-EPI-VASO unveils task activation located in the cortical laminae with little contamination from surface veins, in contrast to the cortical surface weighting of GE-BOLD fMRI. In the physiological-noise-dominated regime of lower resolutions, however, 2D-SMS-VASO shows superior performance compared to 3D-EPI-VASO. Due to its superior sensitivity at a layer-dependent level, 3D-EPI VASO promises to play an important role in future neuroscientific applications of layer-dependent fMRI.

Model-based iterative reconstruction for single-shot EPI at 7T

PURPOSE

To describe a model-based reconstruction strategy for single-shot echo planar imaging (EPI) that intrinsically accounts for k-space nonuniformity, Nyquist ghosting, and geometric distortions during rather than before or after image reconstruction.

METHODS

Ramp sampling and inhomogeneous B0 field-induced distortion cause the EPI samples to lie on a non-Cartesian grid, thus requiring the nonuniform fast Fourier transform. Additionally, a 2D Nyquist ghost phase correction without the need for extra navigator acquisition is included in the proposed reconstruction. Coil compression is also incorporated to reduce the computational load. The proposed method is applied to phantom and human brain MRI data.

RESULTS

The results demonstrate that Nyquist ghosting and geometric distortions are reduced by the proposed reconstruction. The proposed 2D phase correction is superior to a conventional 1D correction. The reductions of both artifacts lead to improved temporal signal-to-noise ratio (tSNR). The virtual coil results suggest that the processing time can be reduced by up to 75%, with a mean tSNR loss of only 3.2% when using 8-virtual instead of 32-physical coils for twofold undersampled data.

CONCLUSION

The proposed reconstruction improves the quality (ghosting, geometry, and tSNR) of EPI without requiring calibration data for Nyquist ghost correction. Magn Reson Med 78:2250-2264, 2017. © 2017 International Society for Magnetic Resonance in Medicine.

Reduction of across-run variability of temporal SNR in accelerated EPI time-series data through FLEET-based robust autocalibration

Temporal signal-to-noise ratio (tSNR) is a key metric for assessing the ability to detect brain activation in fMRI data. A recent study has shown substantial variation of tSNR between multiple runs of accelerated EPI acquisitions reconstructed with the GRAPPA method using protocols commonly used for fMRI experiments. Across-run changes in the location of high-tSNR regions could lead to misinterpretation of the observed brain activation patterns, reduced sensitivity of the fMRI studies, and biased results. We compared conventional EPI autocalibration (ACS) methods with the recently-introduced FLEET ACS method, measuring their tSNR variability, as well as spatial overlap and displacement of high-tSNR clusters across runs in datasets acquired from human subjects at 7T and 3T. FLEET ACS reconstructed data had higher tSNR levels, as previously reported, as well as better temporal consistency and larger overlap of the high-tSNR clusters across runs compared with reconstructions using conventional multi-shot (ms) EPI ACS data. tSNR variability across two different runs of the same protocol using ms-EPI ACS data was about two times larger than for the protocol using FLEET ACS for acceleration factors (R) 2 and 3, and one and half times larger for R=4. The level of across-run tSNR consistency for data reconstructed with FLEET ACS was similar to within-run tSNR consistency. The displacement of high-tSNR clusters across two runs (inter-cluster distance) decreased from ∼8mm in the time-series reconstructed using conventional ms-EPI ACS data to ∼4mm for images reconstructed using FLEET ACS. However, the performance gap between conventional ms-EPI ACS and FLEET ACS narrowed with increasing parallel imaging acceleration factor. Overall, the FLEET ACS method provides a simple solution to the problem of varying tSNR across runs, and therefore helps ensure that an assumption of fMRI analysis-that tSNR is largely consistent across runs-is met for accelerated acquisitions.

Improved adaptive reconstruction of multichannel MR images

PURPOSE

To improve adaptive reconstruction of multichannel MR images by simultaneously removing nonsmooth phase and signal-loss imaging artifacts.

METHODS

The improved adaptive reconstruction consists of three steps: (1) modified multichannel images are first derived by dividing raw multichannel images by a reference image (i.e., a normalized single-channel image); (2) the modified multichannel images are smoothed by a low-pass filter; (3) adaptive spatial matched filters determined from the smoothed multichannel images are utilized to obtain multichannel combined images. Numerical simulations, as well as MRI experiments, on phantoms and human subjects are performed to evaluate and compare the effectiveness of this improved adaptive reconstruction approach against traditional coil combination methods.

RESULTS

Both simulation and MRI experimental results demonstrated that the proposed improved adaptive reconstruction method is able to obtain combined images with reduced nonsmooth phase and signal-loss imaging artifacts.

CONCLUSIONS

A novel multichannel image reconstruction method is developed that produces high quality multichannel combined images.

Tradeoffs in pushing the spatial resolution of fMRI for the 7T Human Connectome Project

Whole-brain functional magnetic resonance imaging (fMRI), in conjunction with multiband acceleration, has played an important role in mapping the functional connectivity throughout the entire brain with both high temporal and spatial resolution. Ultrahigh magnetic field strengths (7T and above) allow functional imaging with even higher functional contrast-to-noise ratios for improved spatial resolution and specificity compared to traditional field strengths (1.5T and 3T). High-resolution 7T fMRI, however, has primarily been constrained to smaller brain regions given the amount of time it takes to acquire the number of slices necessary for high resolution whole brain imaging. Here we evaluate a range of whole-brain high-resolution resting state fMRI protocols (0.9, 1.25, 1.5, 1.6 and 2mm isotropic voxels) at 7T, obtained with both in-plane and slice acceleration parallel imaging techniques to maintain the temporal resolution and brain coverage typically acquired at 3T. Using the processing pipeline developed by the Human Connectome Project, we demonstrate that high resolution images acquired at 7T provide increased functional contrast to noise ratios with significantly less partial volume effects and more distinct spatial features, potentially allowing for robust individual subject parcellations and descriptions of fine-scaled patterns, such as visuotopic organization.

Rapid brain MRI acquisition techniques at ultra-high fields

Ultra-high-field MRI provides large increases in signal-to-noise ratio (SNR) as well as enhancement of several contrast mechanisms in both structural and functional imaging. Combined, these gains result in a substantial boost in contrast-to-noise ratio that can be exploited for higher-spatial-resolution imaging to extract finer-scale information about the brain. With increased spatial resolution, however, there is a concurrent increased image-encoding burden that can cause unacceptably long scan times for structural imaging and slow temporal sampling of the hemodynamic response in functional MRI - particularly when whole-brain imaging is desired. To address this issue, new directions of imaging technology development - such as the move from conventional 2D slice-by-slice imaging to more efficient simultaneous multislice (SMS) or multiband imaging (which can be viewed as "pseudo-3D" encoding) as well as full 3D imaging - have provided dramatic improvements in acquisition speed. Such imaging paradigms provide higher SNR efficiency as well as improved encoding efficiency. Moreover, SMS and 3D imaging can make better use of coil sensitivity information in multichannel receiver arrays used for parallel imaging acquisitions through controlled aliasing in multiple spatial directions. This has enabled unprecedented acceleration factors of an order of magnitude or higher in these imaging acquisition schemes, with low image artifact levels and high SNR. Here we review the latest developments of SMS and 3D imaging methods and related technologies at ultra-high field for rapid high-resolution functional and structural imaging of the brain. Copyright © 2016 John Wiley & Sons, Ltd.

High-resolution diffusion MRI at 7T using a three-dimensional multi-slab acquisition

High-resolution diffusion MRI can provide the ability to resolve small brain structures, enabling investigations of detailed white matter architecture. A major challenge for in vivo high-resolution diffusion MRI is the low signal-to-noise ratio. In this work, we combine two highly compatible methods, ultra-high field and three-dimensional multi-slab acquisition to improve the SNR of high-resolution diffusion MRI. As each k_z plane is encoded using a single-shot echo planar readout, scan speeds of the proposed technique are similar to the commonly used two-dimensional diffusion MRI. In-plane parallel acceleration is applied to reduce image distortions. To reduce the sensitivity of auto-calibration signal data to subject motion and respiration, several new adaptions of the fast low angle excitation echo-planar technique (FLEET) that are suitable for 3D multi-slab echo planar imaging are proposed and evaluated. A modified reconstruction scheme is proposed for auto-calibration with the most robust method, Slice-FLEET acquisition, to make it compatible with navigator correction of motion induced phase errors. Slab boundary artefacts are corrected using the nonlinear slab profile encoding method recently proposed by our group. In vivo results demonstrate that using 7T and three-dimensional multi-slab acquisition with improved auto-calibration signal acquisition and nonlinear slab boundary artefacts correction, high-quality diffusion MRI data with ~1mm isotropic resolution can be achieved.

Reducing sensitivity losses due to respiration and motion in accelerated echo planar imaging by reordering the autocalibration data acquisition

PURPOSE

To reduce the sensitivity of echo-planar imaging (EPI) auto-calibration signal (ACS) data to patient respiration and motion to improve the image quality and temporal signal-to-noise ratio (tSNR) of accelerated EPI time-series data.

METHODS

ACS data for accelerated EPI are generally acquired using segmented, multishot EPI to distortion-match the ACS and time-series data. The ACS data are, therefore, typically collected over multiple TR periods, leading to increased vulnerability to motion and dynamic B0 changes. The fast low-angle excitation echo-planar technique (FLEET) is adopted to reorder the ACS segments so that segments within any given slice are acquired consecutively in time, thereby acquiring ACS data for each slice as rapidly as possible.

RESULTS

Subject breathhold and motion phantom experiments demonstrate that artifacts in the ACS data reduce tSNR and produce tSNR discontinuities across slices in the accelerated EPI time-series data. Accelerated EPI data reconstructed using FLEET-ACS exhibit improved tSNR and increased tSNR continuity across slices. Additionally, image quality is improved dramatically when bulk motion occurs during the ACS acquisition.

CONCLUSION

FLEET-ACS provides reduced respiration and motion sensitivity in accelerated EPI, which yields higher tSNR and image quality. Benefits are demonstrated in both conventional-resolution 3T and high-resolution 7T EPI time-series data.

High resolution whole brain diffusion imaging at 7T for the Human Connectome Project

Mapping structural connectivity in healthy adults for the Human Connectome Project (HCP) benefits from high quality, high resolution, multiband (MB)-accelerated whole brain diffusion MRI (dMRI). Acquiring such data at ultrahigh fields (7T and above) can improve intrinsic signal-to-noise ratio (SNR), but suffers from shorter T2 and T2(⁎) relaxation times, increased B1(+) inhomogeneity (resulting in signal loss in cerebellar and temporal lobe regions), and increased power deposition (i.e. specific absorption rate (SAR)), thereby limiting our ability to reduce the repetition time (TR). Here, we present recent developments and optimizations in 7T image acquisitions for the HCP that allow us to efficiently obtain high quality, high resolution whole brain in-vivo dMRI data at 7T. These data show spatial details typically seen only in ex-vivo studies and complement already very high quality 3T HCP data in the same subjects. The advances are the result of intensive pilot studies aimed at mitigating the limitations of dMRI at 7T. The data quality and methods described here are representative of the datasets that will be made freely available to the community in 2015.

Improvement of temporal signal-to-noise ratio of GRAPPA accelerated echo planar imaging using a FLASH based calibration scan

PURPOSE

To demonstrate that the temporal signal-to-noise ratio (SNR) of generalized autocalibrating partially parallel acquisitions (GRAPPA) accelerated echo planar imaging (EPI) can be enhanced and made more spatially uniform by using a fast low angle shot (FLASH) based calibration scan.

METHODS

EPI of a phantom and human brains were acquired at 3 Tesla without and with GRAPPA acceleration factor of 2. The GRAPPA accelerated data were reconstructed using calibration scans acquired with EPI and FLASH acquisition schemes. The increase in temporal signal fluctuation due to GRAPPA reconstruction was quantified and compared. Simulated g-factor maps were also created for different calibration scans.

RESULTS

GRAPPA accelerated phantom data exhibited areas with high g values when using the EPI based calibration for reconstruction. The g-factor maps were uniform when using the FLASH calibration scan. g was greater than 1.1 in 74% of pixels in 64 × 64 data reconstructed with the EPI calibration compared with only 15% when using the FLASH calibration scan. Human data also showed abnormally high g regions when using the EPI calibration but not when using the FLASH calibration scan. Use of the FLASH calibration scan increased the whole brain temporal SNR by ∼12% without affecting the image quality. Experimental observations were confirmed by simulations.

CONCLUSION

A calibration scan based on a FLASH acquisition scheme can be used to improve the temporal SNR of GRAPPA accelerated EPI time series. Magn Reson Med 75:2362-2371, 2016. Published 2015. This article is a U.S. Government work and is in the public domain in the USA.

Advances in real-time phase-contrast flow MRI using asymmetric radial gradient echoes

PURPOSE

To provide multidimensional velocity compensation for real-time phase-contrast flow MRI.

METHODS

The proposed method introduces asymmetric gradient echoes for highly undersampled radial FLASH MRI with phase-sensitive image reconstruction by regularized nonlinear inversion (NLINV). Using an adapted gradient delay correction the resulting image quality was analyzed by simulations and experimentally validated at 3 Tesla. For real-time flow MRI the reduced gradient-echo timing allowed for the incorporation of velocity-compensating waveforms for all imaging gradients at even shorter repetition times.

RESULTS

The results reveal a usable degree of 20% asymmetry. Real-time flow MRI with full velocity compensation eliminated signal void in a flow phantom, confirmed flow parameters in healthy subjects and demonstrated signal recovery and phase conservation in a patient with aortic valve insufficiency and stenosis. Exemplary protocols at 1.4-1.5 mm resolution and 6 mm slice thickness achieved total acquisition times of 33.3-35.7 ms for two images (7 spokes each) with and without flow-encoding gradient.

CONCLUSION

Asymmetric gradient echoes were successfully implemented for highly undersampled radial trajectories. The resulting temporal gain offers full velocity compensation for real-time phase-contrast flow MRI which minimizes false-positive contributions from complex flow and further enhances the temporal resolution compared with acquisitions with symmetric echoes.

Motion robust GRAPPA for echo-planar imaging

PURPOSE

A readout-segmented method for acquiring robust GRAPPA calibration data for echo-planar imaging (EPI) was proposed and compared with two previous methods, including the gold standard interleaved approach and a single shot method with halved phase encode resolution.

THEORY AND METHODS

The readout-segmented and single shot techniques acquire adjacent phase encode lines in the same shot to obtain the calibration data, rather than interleaving of lines between shots, to decrease sensitivity to motion. Additionally, it uses multiple segments with shortened frequency encode extent to match the phase encode bandwidth to the undersampled data, which decreases sensitivity to B0 inhomogeneity. The three methods were tested using simulations and EPI scans of the brain in healthy volunteers.

RESULTS

The interleaved approach exhibited high sensitivity to motion, while residual undersampling artifacts remained in the single shot method due to mismatch of B0 inhomogeneity between the calibration and undersampled data. The readout segmented method exhibited no such errors, having 30% lower ghosting intensity than the single shot method and 90% lower ghosting intensity than the interleaved approach in moving subjects.

CONCLUSION

Artifacts from B0 inhomogeneity and motion during calibration scans for EPI GRAPPA can be mitigated with a readout segmented calibration scan.

Quantifying the statistical impact of GRAPPA in fcMRI data with a real-valued isomorphism

The interpolation of missing spatial frequencies through the generalized auto-calibrating partially parallel acquisitions (GRAPPA) parallel magnetic resonance imaging (MRI) model implies a correlation is induced between the acquired and reconstructed frequency measurements. As the parallel image reconstruction algorithms in many medical MRI scanners are based on the GRAPPA model, this study aims to quantify the statistical implications that the GRAPPA model has in functional connectivity studies. The linear mathematical framework derived in the work of Rowe , 2007, is adapted to represent the complex-valued GRAPPA image reconstruction operation in terms of a real-valued isomorphism, and a statistical analysis is performed on the effects that the GRAPPA operation has on reconstructed voxel means and correlations. The interpolation of missing spatial frequencies with the GRAPPA model is shown to result in an artificial correlation induced between voxels in the reconstructed images, and these artificial correlations are shown to reside in the low temporal frequency spectrum commonly associated with functional connectivity. Through a real-valued isomorphism, such as the one outlined in this manuscript, the exact artificial correlations induced by the GRAPPA model are not simply estimated, as they would be with simulations, but are precisely quantified. If these correlations are unaccounted for, they can incur an increase in false positives in functional connectivity studies.

Analysis of proton-density bias corrections based on T1 measurement for robust quantification of water content in the brain at 3 Tesla

PURPOSE

Estimating tissue water content using high field MRI, such as 3 Tesla (T), is challenging due to the difficulty in dissociating the radio frequency inhomogeneity pattern from the signal arising from tissue intrinsic proton density (PD) variations. To overcome this problem the longitudinal relaxation time T1 can be combined with an initial guess of the PD to yield the desired PD bias correction. However, it is necessary to know whether T1 effects, i.e., any effect contributing to T1 while being independent of tissue hydration, influence the estimated correction.

METHODS

Twenty-five healthy subjects underwent a quantitative 3T MRI protocol enabling acquisition of 64 slices with 1 mm in-plane resolution and 2 mm slice thickness in 14 min. Influence of T1 effects on the estimated water content map is evaluated using a dedicated method including T1 and T2 * information and region of interest-based water content values are compared with the literature.

RESULTS

Our analysis indicates that the PD bias correction based on T1 is largely insensitive to T1 effects. Besides, water content results are in good agreement with literature values obtained at 1.5T.

CONCLUSION

This study demonstrates the applicability of a PD bias correction based on T1 to yield tissue water content at 3T.

Nonlinear coil sensitivity estimation for parallel magnetic resonance imaging using data-adaptive steering kernel regression method

The parallel magnetic resonance imaging (parallel imaging) technique reduces the MR data acquisition time by using multiple receiver coils. Coil sensitivity estimation is critical for the performance of parallel imaging reconstruction. Currently, most coil sensitivity estimation methods are based on linear interpolation techniques. Such methods may result in Gibbs-ringing artifact or resolution loss, when the resolution of coil sensitivity data is limited. To solve the problem, we proposed a nonlinear coil sensitivity estimation method based on steering kernel regression, which performs a local gradient guided interpolation to the coil sensitivity. The in vivo experimental results demonstrate that this method can effectively suppress Gibbs ringing artifact in coil sensitivity and reduces both noise and residual aliasing artifact level in SENSE reconstruction.

SENSE with improved tolerance to inaccuracies in coil sensitivity maps

In this work, an extension of the Cartesian sensitivity encoding (SENSE) parallel imaging framework is proposed. In the well-known SENSE solution, the overdetermined reconstruction inversion problem is optimized to get the highest signal-to-noise ratio in the image. In this extension, the probability of artifacts due to incorrect knowledge of the receiver coil sensitivities is also taken into account. This is realized by assuming an uncertainty in measured receiver coil sensitivities to enable weighting of residual artifact level and signal-to-noise ratio in the inversion problem. This inversion problem can still be solved by a least-squares optimization without the need of any complex iterative scheme. Results in abdominal imaging show that artifact levels can be substantially reduced, at the cost of a signal-to-noise ratio penalty. The size of the signal-to-noise ratio penalty depends on the assumed inaccuracy of the coil sensitivities, sensitivity encoding acceleration factor, and coil configuration.

Parallel and partial Fourier imaging with prospective motion correction

Subject motion during scan is a major source of artifacts in MR examinations. Prospective motion correction is a promising technique that tracks subject motion and adjusts the imaging volume in real time; however, additional retrospective correction may be necessary to achieve robust image quality and compatibility with other imaging options. Real-time realignment of the imaging volume by prospective motion correction changes the coil sensitivity weighting and the field inhomogeneity relative to the imaging volume. This can pose image reconstruction problems with parallel imaging and partial Fourier imaging, which rely on coil sensitivity and image phase information, respectively. This work presents a practical method for reconstructing images acquired using prospective motion correction with parallel imaging and/or partial Fourier imaging. Our proposed approach is data driven and noniterative; data are binned into several position bins based on motion measurements made during the prospective motion correction acquisition and the data in each bin are processed through intrabin operations such as parallel imaging reconstruction (in case of undersampling), phase correction, and coil combination before combination of the position bins. We demonstrate the effectiveness of our technique through simulation studies and in vivo experiments using a prospectively motion-corrected three-dimensional fast spin echo sequence.

Parallel reconstruction using null operations

A novel iterative k-space data-driven technique, namely parallel reconstruction using null operations (PRUNO), is presented for parallel imaging reconstruction. In PRUNO, both data calibration and image reconstruction are formulated into linear algebra problems based on a generalized system model. An optimal data calibration strategy is demonstrated by using singular value decomposition, and an iterative conjugate-gradient approach is proposed to efficiently solve missing k-space samples during reconstruction. With its generalized formulation and precise mathematical model, PRUNO reconstruction yields good accuracy, flexibility, and stability. Both computer simulation and in vivo studies have shown that PRUNO produces much better reconstruction quality than generalized autocalibrating partially parallel acquisition (GRAPPA), especially under high accelerating rates. With the aid of PRUNO reconstruction, ultra-high accelerating parallel imaging can be performed with decent image quality. For example, we have done successful PRUNO reconstruction at a reduction factor of 6 (effective factor of 4.44) with eight coils and only a few autocalibration signal lines.

Derivative encoding for parallel magnetic resonance imaging

PURPOSE

To introduce a linear shift-invariant relationship between the partial derivatives of k space signals acquired using multichannel receive coils and to demonstrate that k space derivatives can be used for image unwrapping.

METHODS

Fourier transform of k space derivatives contains information on the spatial origins of aliased pixels; therefore, images can be reconstructed by k space derivatives. Fully sampled phantom and brain images acquired at 3 T using a standard eight channel receive coil were used to validate the k space derivatives theorem by unwrapping aliased images.

RESULTS

Derivative encoding leads to new methods for parallel imaging reconstruction in both k space and image domains. Noise amplification in sensitivity encoding image reconstruction, which is considered to produce the optimal SNR, can be further reduced using k space derivative encoding without making any assumptions on the characteristics of the images to be reconstructed.

CONCLUSIONS

This work demonstrated that the partial derivative of the k space signal acquired from one coil with respect to one direction can be expressed as a sum of partial derivatives of signals from multiple coils with respect to the perpendicular k space direction(s). This relationship between the partial derivatives of k space signals is linear and shift-invariant in the Cartesian coordinate system.

Contrast-enhanced whole-heart coronary magnetic resonance angiography at 3 T using interleaved echo planar imaging

OBJECTIVES

The goal of this work was to reduce the scan time of contrast-enhanced whole-heart coronary magnetic resonance angiography (MRA) by using a gradient echo interleaved echo planar imaging (GRE-EPI) sequence at 3 T field strength.

MATERIALS AND METHODS

A GRE-EPI sequence was optimized to acquire contrast-enhanced whole-heart coronary MRA at 3 T. First-order phase correction was used for alignment of the odd and even echoes in the GRE-EPI echo train. Single and dual reference scan techniques for estimation of the linear phase correction parameters were evaluated using both phantom and volunteer studies. The GRE-EPI readout was combined with parallel imaging for a further reduction in scan time. To avoid image distortions, calibration signals for coil sensitivity estimation were acquired in a separate low resolution GRE scan before the whole-heart GRE-EPI scan. Eight healthy volunteers were scanned with the optimized contrast-enhanced GRE-EPI sequence. GRE-EPI images were acquired during slow infusion (0.3 mL/s) of 0.1 mmol/kg body weight of Gd-BOPTA. For comparison purposes, the same 8 volunteers were scanned again in a separate scan session using a traditional GRE sequence with double the dose (0.2 mmol/kg body weight) of the same contrast agent with the same injection rate. The contrast-enhanced GRE-EPI and contrast-enhanced GRE techniques were compared in terms of relative signal-to-noise ratio (rSNR), relative contrast-to-noise ratio (rCNR), image quality scores, and visualized vessel lengths.

RESULTS

Both, phantom and volunteer studies demonstrated that the dual reference scan phase correction technique was a key step for obtaining satisfactory image quality using GRE-EPI at 3 T. Whole-heart coronary MRA with a spatial resolution of 1.0 x 1.0 x 2.0 mm3 was acquired with the GRE-EPI sequence in an average scan time of 2.5 +/- 0.6 minutes, compared with 8.6 +/- 2.7 minutes for the GRE technique. The GRE-EPI technique had lower rCNR compared with the GRE sequence. The image quality and coronary artery visualization with the GRE-EPI technique were adequate, and there was no statistically significant difference in the image quality scores, rSNR, and visualized coronary artery lengths between the GRE-EPI and GRE techniques.

CONCLUSIONS

Contrast-enhanced whole-heart coronary MRA using the GRE-EPI technique resulted in excellent delineation of all the major coronary arteries and compared with current GRE techniques demonstrated a factor of 2 reduction in contrast agent dose and a factor of 3 reduction in scan time.

DISTORTION-OPTIMAL SELF-CALIBRATING PARALLEL MRI BY BLIND INTERPOLATION IN SUBSAMPLED FILTER BANKS

Self-calibrating k-space-based image reconstruction in parallel MRI interpolates the subsampled multi-channel data to a fully sampled Nyquist grid in k-space. Adopting a filter bank interpolation framework, we provide a new formulation of the associated inverse problem and develop the theory for blind identification of the interpolant filters. The developed method is applied to imaging scenarios where high effective acceleration is desired and is shown to be capable of reconstructing artifact-free images with minimal amount of calibration data - hence, achieving high effective accelerations. Simulation and in-vivo results indicate that improved image quality, and thus greater scan time reductions compared to the state-of-the-art method of GRAPPA can be achieved.

An electromagnetic reverse method of coil sensitivity mapping for parallel MRI - theoretical framework

In this paper, a novel sensitivity mapping method is proposed for the image domain parallel MRI (pMRI) technique. Instead of refining raw sensitivity maps by means of conventional image processing operations such as polynomial fitting, the presented method determines coil sensitivity profiles through an iterative optimization process. During the algorithm implementation the optimization cost function is defined as the difference between the raw sensitivity profile and the desired profile. The minimization is governed by the physics of low-frequency electromagnetic and reciprocity theories. The performance of the method was theoretically investigated and compared with that of a traditional polynomial fitting, against a range of system noise levels. It was found that, the new method produces high-fidelity sensitivity profiles with noise amplitudes, measured as root mean square deviation an order of magnitude less than that of the polynomial fitting method. Using the sensitivity profiles generated by our method, SENSE (sensitivity encoding) reconstructions produce significantly less image artefacts than conventional methods. The successful implementation of this method has far-reaching implications that accurate sensitivity mapping is not only important for parallel reconstruction, but also essential for its transmission analogy, such as Transmit SENSE.

Increasing efficiency of parallel imaging for 2D multislice acquisitions

Parallel imaging algorithms require precise knowledge about the spatial sensitivity variation of the receiver coils to reconstruct images with full field of view (FOV) from undersampled Fourier encoded data. Sensitivity information must either be given a priori, or estimated from calibration data acquired along with the actual image data. In this study, two approaches are presented, which require very little or no additional data at all for calibration in two-dimensional multislice acquisitions. Instead of additional data, information from spatially adjacent slices is used to estimate coil sensitivity information, thereby increasing the efficiency of parallel imaging. The proposed approaches rely on the assumption that over a small range of slices, coil sensitivities vary smoothly in slice direction. Both methods are implemented as variants of the GRAPPA algorithm. For a given effective acceleration, superior image quality is achieved compared to the conventional GRAPPA method. For the latter calibration lines for coil weight computation must be acquired in addition to the undersampled k-spaces for coil weight computation, thus requiring higher k-space undersampling, that is, a higher reduction factor to achieve the same effective acceleration. The proposed methods are particularly suitable to speed up parallel imaging for clinical applications where the reduction factor is limited to two or three.

Image reconstruction by regularized nonlinear inversion--joint estimation of coil sensitivities and image content

The use of parallel imaging for scan time reduction in MRI faces problems with image degradation when using GRAPPA or SENSE for high acceleration factors. Although an inherent loss of SNR in parallel MRI is inevitable due to the reduced measurement time, the sensitivity to image artifacts that result from severe undersampling can be ameliorated by alternative reconstruction methods. While the introduction of GRAPPA and SENSE extended MRI reconstructions from a simple unitary transformation (Fourier transform) to the inversion of an ill-conditioned linear system, the next logical step is the use of a nonlinear inversion. Here, a respective algorithm based on a Newton-type method with appropriate regularization terms is demonstrated to improve the performance of autocalibrating parallel MRI--mainly due to a better estimation of the coil sensitivity profiles. The approach yields images with considerably reduced artifacts for high acceleration factors and/or a low number of reference lines.

Comparison of parallel acquisition techniques generalized autocalibrating partially parallel acquisitions (GRAPPA) and modified sensitivity encoding (mSENSE) in functional MRI (fMRI) at 3T

PURPOSE

To evaluate the parallel acquisition techniques, generalized autocalibrating partially parallel acquisitions (GRAPPA) and modified sensitivity encoding (mSENSE), and determine imaging parameters maximizing sensitivity toward functional activation at 3T.

MATERIALS AND METHODS

A total of eight imaging protocols with different parallel imaging techniques (GRAPPA and mSENSE) and reduction factors (R = 1, 2, 3) were compared at different matrix sizes (64 and 128) with respect to temporal noise characteristics, artifact behavior, and sensitivity toward functional activation.

RESULTS

Echo planar imaging (EPI) with GRAPPA and a reduction factor of 2 revealed similar image quality and sensitivity than full k-space EPI. A higher incidence of artifacts and a marked sensitivity loss occurred at R = 3. Even though the same eight-channel head coil was used for signal detection in all experiments, GRAPPA generally showed more benign patterns of spatially-varying noise amplification, and mSENSE was also more susceptible to residual unfolding artifacts than GRAPPA.

CONCLUSION

At 3T and a reduction factor of 2, parallel imaging can be used with only little penalty with regard to sensitivity. With our implementation and coil setup the performance of GRAPPA was clearly superior to mSENSE. Thus, it seems advisable to pay special attention to the employed parallel imaging method and its implementation.

Image quality improvement of composed whole-spine MR images by applying a modified homomorphic filter--first results in cases of multiple myeloma

To establish a modified homomorphic filter (BiFiC) for post-processing of composed MR images in clinical routine and to evaluate it in special regards to image quality and diagnostic safety. Twenty-three whole-spine examinations were post-processed with the filter. Qualitative image evaluation included documentation of lesions and their visualization at original and post-processed images. Variations of signal intensities were calculated pixel by pixel and visualized by color-coded maps. Quantitative data evaluation was conducted by region-by-region analysis with standardized regions of interests. The BiFiC filter could be implemented successfully on the scanner's software platform and used within clinical routine. Color-coded maps could demonstrate that the BiFiC filter improves the signal uniformity in all cases, including images with metallic artifacts caused by implants. The subjective image quality of the post-processed images was improved in 22 out of the 23 MR examinations; in one case it was rated as equal. All pathologies were visualized on post-processed images without the need of additional contrast adjustments. The implemented BiFiC filter significantly improves image signal homogeneity. The algorithm can consequently be integrated into clinical routine as an automatic image post-processing step.

Comparison of reconstruction accuracy and efficiency among autocalibrating data-driven parallel imaging methods

The class of autocalibrating "data-driven" parallel imaging (PI) methods has gained attention in recent years due to its ability to achieve high quality reconstructions even under challenging imaging conditions. The aim of this work was to perform a formal comparative study of various data-driven reconstruction techniques to evaluate their relative merits for certain imaging applications. A total of five different reconstruction methods are presented within a consistent theoretical framework and experimentally compared in terms of the specific measures of reconstruction accuracy and efficiency using one-dimensional (1D)-accelerated Cartesian datasets. It is shown that by treating the reconstruction process as two discrete phases, a calibration phase and a synthesis phase, the reconstruction pathway can be tailored to exploit the computational advantages available in certain data domains. A new "split-domain" reconstruction method is presented that performs the calibration phase in k-space (k(x), k(y)) and the synthesis phase in a hybrid (x, k(y)) space, enabling highly accurate 2D neighborhood reconstructions to be performed more efficiently than previously possible with conventional techniques. This analysis may help guide the selection of PI methods for a given imaging task to achieve high reconstruction accuracy at minimal computational expense.

Parallel imaging reconstruction for arbitrary trajectories using k-space sparse matrices (kSPA)

Although the concept of receiving MR signal using multiple coils simultaneously has been known for over two decades, the technique has only recently become clinically available as a result of the development of several effective parallel imaging reconstruction algorithms. Despite the success of these algorithms, it remains a challenge in many applications to rapidly and reliably reconstruct an image from partially-acquired general non-Cartesian k-space data. Such applications include, for example, three-dimensional (3D) imaging, functional MRI (fMRI), perfusion-weighted imaging, and diffusion tensor imaging (DTI), in which a large number of images have to be reconstructed. In this work, a systematic k-space-based reconstruction algorithm based on k-space sparse matrices (kSPA) is introduced. This algorithm formulates the image reconstruction problem as a system of sparse linear equations in k-space. The inversion of this system of equations is achieved by computing a sparse approximate inverse matrix. The algorithm is demonstrated using both simulated and in vivo data, and the resulting image quality is comparable to that of the iterative sensitivity encoding (SENSE) algorithm. The kSPA algorithm is noniterative and the computed sparse approximate inverse can be applied repetitively to reconstruct all subsequent images. This algorithm, therefore, is particularly suitable for the aforementioned applications.