Parallel MRI with extended and averaged GRAPPA kernels (PEAK‐GRAPPA): Optimized spatiotemporal dynamic imaging

Abstract

Purpose:To evaluate an optimized k-t-space related reconstruction method for dynamic magnetic resonance imaging (MRI), a method called PEAK-GRAPPA (Parallel MRI with Extended and Averaged GRAPPA Kernels) is presented which is based on an extended spatiotemporal GRAPPA kernel in combination with temporal averaging of coil weights. Materials and Methods:The PEAK-GRAPPA kernel consists of a uniform geometry with several spatial and temporal source points from acquired k-space lines and several target points from missing k-space lines. In order to improve the quality of coil weight estimation sets of coil weights are averaged over the temporal dimension. Results:The kernel geometry leads to strongly decreased reconstruction times compared to the recently introduced k-t-GRAPPA using different kernel geometries with only one target point per kernel to fit. Improved results were obtained in terms of the root mean square error and the signal-to-noise ratio as demonstrated by in vivo cardiac imaging. Conclusion:Using a uniform kernel geometry for weight estimation with the properties of uncorrelated noise of different acquired timeframes, optimized results were achieved in terms of error level, signal-to-noise ratio, and reconstruction time. DYNAMIC MRI is an important foundation for many clinical applications such as time-resolved (Cine) cardiac imaging for the assessment of left ventricular function. To achieve sufficient spatial and temporal resolution fast data acquisition is essential, particularly for applications that require breath-holding. In order to reduce total acquisition time or to increase spatiotemporal resolution, parallel imaging techniques such as SENSE or GRAPPA have been introduced. By decreasing the number of phase-encoding steps by a reduction factor R, imaging can be substantially accelerated (1,2). Based on varying sensitivity in multiple receiver coil arrays, parallel imaging reconstruction algorithms remove resulting aliasing artifacts either in the image domain (2) or regenerate the missing data in k-space (1). For parallel imaging reconstruction using the kspace related GRAPPA technique the image reconstruction and the combination of images from different receiver coils are decoupled. Therefore, the process of unaliasing of the uncombined coil images (generated from each coil) can be optimized separately.For conventional parallel dynamic MRI using GRAPPA, the central k-space for each timeframe is fully sampled, forming the autocalibration signal (ACS) lines, while the outer k-space is undersampled in the phase-encoding (ky) direction according to a user-defined reduction factor R. All timeframes are reconstructed independently using a kernel with a certain extension in kx-and ky-direction. The kernel is shifted across the ACS lines in order to estimate the coil weights needed for the reconstruction (or interpolation) of the missing lines in outer k-space. For the reconstruction process of the missing k-space lines, the kernel is shifted by an increment of R in ky-direction over the undersampl...

k‐t‐space accelerated myocardial perfusion

“…SNR was determined by dividing the mean value within the ROI of the averaged image by the standard deviation in the identical ROI in the subtracted image. Only pairs of time frames with a temporal distance of the reduction factor R ϭ 6 were used for SNR calculation because time frames reconstructed within the target boundary of a kernel show an influence on the SNR due to an introduction of temporally correlated noise (10). Time frames 2 and 8 were chosen for SNR calculation to use frames before the arrival of the contrast agent bolus.…”

“…The observed SNR optimization is a result of the inclusion of temporal information for the estimation of coil weights and reconstruction of missing k-space data. Moreover, GRAPPA weight averaging effectively exploits temporally uncorrelated noise in different time frames and results in optimized SNR performance compared to other parallel imaging techniques (10).…”

“…Recently, the PEAK-GRAPPA spatiotemporal parallel imaging technique was introduced (10). PEAK-GRAPPA is based on an extended spatiotemporal uniform GRAPPA kernel geometry in combination with temporal averaging of sets of coil weights obtained over k-t space.…”

“…PEAK-GRAPPA is based on an extended spatiotemporal uniform GRAPPA kernel geometry in combination with temporal averaging of sets of coil weights obtained over k-t space. The use of such a uniform kernel geometry leads to a decreased artifact level, as well as to strongly decreased reconstruction times (10).…”

Fast T2 relaxometry with an accelerated multi‐echo spin‐echo sequence

“…There are also efforts performed that exploit correlations in both k-space and time space (24)(25)(26)(27)(28), denoted as k-t-space thereafter. The advantage of the latter approach is that it exploits more relevant information, thus improving the estimation of the missing data.…”