Local reconstruction of stenosed sections of artery using multiple MRA acquisitions

Abstract

A method for reconstructing magnetic resonance angiography (MRA) volumes from successive acquisitions is described. The method is based on double oblique acquisitions of highly anisotropic MRA volumes, each of which corresponds to reduced k-space filling. These partial k-spaces are then combined to obtain a 3D k-space adapted to the frequency spread of the angiographic image of the stenosis. The SNR-resolution compromise of MRA is thus improved by focusing the acquisition on the most relevant k-space regions. The reconstruction is performed directly in k-space by averaging the partial k-spaces. The characteristics of MR images are determined by a trade-off between three factors-the voxel size, the acquisition time, and the signal-to-noise ratio (SNR)-regardless of the acquisition sequence used. It is particularly difficult to obtain an optimal compromise between these parameters in 3D magnetic resonance angiography (MRA) because the acquisition time is limited by the recirculation of the contrast bolus and the breath-hold capacity of the patient.The acquisition time required to record a whole k-space has been reduced by the use of intense gradients and fast multi-shot or single-shot spin-echo or gradient-echo acquisition sequences (1). Because they have a shorter sampling time, multishot methods (2,3) are less sensitive to signal variations such as chemical shift dephasing and decreased T 2 or T* 2 , and are thus less prone to image distortions. Non-Cartesian filling of k-space has also been used to shorten acquisition time. Radial (4 -6), circular (7,8), stochastic (9), and rosette (10) k-space trajectories have also been proposed. Interleaved spiral acquisitions, which allow a more efficient exploitation of the slew rate of gradient amplifiers, and reduce sensitivity to the magnetic field inhomogeneities, remain the most common choice in cardiac ultrafast imaging. However, these methods call for sophisticated reconstruction techniques, including resampling (5,11,12), gridding (13-15), and least-squares-based interpolation (16 -18). Simultaneous acquisition of spatial harmonics (SMASH) (19,20) and sensitivity encoding (SENSE) (21,22) methods have recently been shown to further reduce acquisition time by providing a parallel acquisition of MR data using phased-array antennas. When combined with fast acquisition sequences, these methods also allow correction for excessive losses in T 2 and T* 2 .Methods that use segmented updating of k-space can also provide time-resolved imaging by "view sharing" after rapid acquisition of multiple 3D volumes throughout the passage of the contrast agent bolus. The Keyhole technique (23), in which a complete k-space is acquired but only the central region is updated, has been improved to optimize k-space filling as a function of the frequency content of the image (24). The block regional interpolation scheme for k-space (BRISK) method (25,26), which uses the continuous acquisition of selected regions of k-space with a strategy based on cardiac dynamic information, was developed f...

Single-image super-resolution of brain MR images using overcomplete dictionaries

“…Two kinds of approaches can be identified: one works at the acquisition level over raw data (frequency space), while the others act on the volumetric images (spatial or image space) as an additional processing step. At the acquisition stage, the k-space data can be manipulated and combined to obtain adequate spatial resolution while reducing acquisition time (Herment et al, 2003); or parameters can be configured to obtain multiple scans with different slice directions which are then mixed up (Shilling et al, 2009). Regarding volumetric images, Peled and Yeshurun (2001) and Greenspan et al (2002) have proposed the first approaches to adapt the iterative back-projection method proposed by Irani and Peleg (1993) to 2D and 3D MR images, respectively; followed by other strategies such as the resolution enhancement method described by Carmi et al (2006).…”

Super‐resolution reconstruction in frequency, image, and wavelet domains to reduce through‐plane partial voluming in MRI

“…When multiple 2D scans with different slice select directions are available, the k-space data from those scans can be combined to provide a dense sampling of the k-space cube involving high-frequency k-space samples in different directions. Examples of these multiple 2D scans are orthogonal acquisitions with axial, coronal, and sagittal slice select directions, 3,9 and acquisitions with rotated slice select directions. 19 The fusion of k-space data from these scans provides dense sampling of the center of k-space and covers high-frequency corners of the k-space as well, thus simultaneously improves SNR and spatial resolution in 3D.…”

“…There has been an interesting body of work on postacquisition fusion of thick-slice anisotropic 2D MRI scans. This includes simple averaging of volumes, selective combination in the Fourier domain, 3 wavelet fusion, 4 and super-resolution reconstruction (SRR) in the image domain. [5][6][7] Super-resolution MRI closely follows the concepts of super-resolution reconstruction in digital image and video processing with techniques in frequency, image, and wavelet domains.…”

7T‐guided super‐resolution of 3T MRI

“…Some other researchers (Greenspan et al, 12 Peeters et al, 13 and Kim et al 14 ) proposed methods to reduce thickness in the slice direction of volumetric images. Other methods (Herment et al, 15 Shilling et al, 16 Carmi et al 17 ) combined multiple shifted scans to achieve a HR result. Woo et al 18 proposed an interpolation method for enhancement of tongue by generating a 3D image volume using three orthogonal images.…”