k‐t BLAST and k‐t SENSE: Dynamic MRI with high frame rate exploiting spatiotemporal correlations

Abstract

Dynamic images of natural objects exhibit significant correlations in k-space and time. Thus, it is feasible to acquire only a reduced amount of data and recover the missing portion afterwards. This leads to an improved temporal resolution, or an improved spatial resolution for a given amount of acquisition. Based on this approach, two methods were developed to significantly improve the performance of dynamic imaging, named k-t BLAST (Broad-use Linear Acquisition Speed-up Technique) and k-t SENSE (SENSitivity Encoding) for use with a single or multiple receiver coils, respectively. Signal correlations were learned from a small set of training data and the missing data were recovered using all available information in a consistent and integral manner. The general theory of k-t BLAST and k-t SENSE is applicable to arbitrary k-space trajectories, timevarying coil sensitivities, and under-and overdetermined reconstruction problems. Examples from ungated cardiac imaging demonstrate a 4-fold acceleration (voxel size 2.42 ؋ 2.52 mm 2 , 38.4 fps) with either one or six receiver coils. k-t BLAST and k-t SENSE are applicable to many areas, especially those exhibiting quasiperiodic motion, such as imaging of the heart, the lungs, the abdomen, and the brain under periodic stimulation. Key words: MRI; fast dynamic imaging; k-t BLAST; k-t SENSE; prior-information-driven parallel imagingDynamic MRI captures an object in motion by acquiring a series of images at a high frame rate. Conceptually, the straightforward approach would be to acquire the full data for reconstructing each time frame separately. This requires the acquisition of each time frame to be short relative to the object motion in order to effectively obtain an instantaneous snapshot. However, this approach is limited by physical (e.g., gradient strength and slew rate) and physiological (e.g., nerve stimulation) constraints on the speed of data acquisition. Over the years a number of strategies have been proposed to further increase the acquisition rate by reducing the amount of acquired data by a given factor, referred to as the acceleration factor hereafter. These strategies are able to reduce data acquisition without compromising image quality significantly because typical image series exhibit a high degree of spatiotemporal correlations, either by nature or by design. Therefore, there is a certain amount of redundancy within the data. In general, such strategies for reducing data acquisition can be divided into three approaches, based on exploiting correlations in k-space, in time, or in both k-space and time.The first approach, based on exploiting correlations in k-space, encompasses a wide variety of methods, including partial Fourier methods (1,2), reduced-field-of-view methods (3), parallel imaging (4,5), and prior-informationdriven methods (1,6). These methods speed up acquisition by collecting only a fraction of k-space at each time frame. The missing data are then recovered based on the measured k-space points from the same time frame. Although the met...

Nonlinear inverse reconstruction for real‐time MRI of the human heart using undersampled radial FLASH

“…As far as the achievable image quality is concerned, the present results may be even further improved by exploiting frame-to-frame temporal redundancies, which are inherently contained in a dynamic image series of the beating heart. Such ideas have previously been shown to improve the reconstruction quality of linear algorithms (13)(14)(15)(16) and will be the subject of future work.…”

Optimized parallel imaging for dynamic PC‐MRI with multidirectional velocity encoding

“…For multidirectional velocity-encoding PC-MRI, ACS lines are typically acquired separately for each velocity direction and each cardiac timeframe. More advanced imaging acceleration techniques such as TGRAPPA (time-interleaved sampling scheme in combination with GRAPPA), TSENSE (adaptive sensitivity encoding incorporating temporal filtering), and several k-t-space related reconstructions, i.e., k-t GRAPPA, PEAK-GRAPPA (parallel MRI with extended and averaged GRAPPA kernels), k-t BLAST (broad-use linear acquisition speed-up technique), and kt SENSE (sensitivity encoding), have recently been reported that exploit the possibility to efficiently undersample both the spatial and temporal dimensions (12)(13)(14)(15)(16). As a result, higher reduction factors can be achieved compared to standard techniques such as GRAPPA or SENSE (5,10).…”

“…1). To evaluate the effect of the number of ACS lines, reduction factors, and retaining central k-lines, images were reconstructed with ACS ¼ [16,32], R ¼ [2,3,4], and [0, 4,8] retained central k-lines (in total, 18 combinations), respectively. As for the true reduction factor of each reconstructed algorithm, the number of used k-line was divided by the number of total k-line of the fully sampled data set.…”

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

“…In TSENSE and TGRAPPA a time-interleaved acquisition scheme allows for a direct merging of adjacent timeframes in order to build a fully encoded reference dataset which is then used to derive sensitivity information or autocalibration signal. In contrast, k-t-SENSE, k-t-BLAST, and k-t-GRAPPA directly incorporate the temporal information into the reconstruction process, thus allowing for higher acceleration factors than available coil elements since these techniques can be treated as a combination of the conventional parallel imaging methods (GRAPPA, SENSE) and sliding window techniques (4,7).…”