Browse Prior Art Database

A PARALLEL IMAGING METHOD USING CYCLIC MATRICES FOR MEDICAL IMAGING SYSTEMS

IP.com Disclosure Number: IPCOM000124163D
Publication Date: 2005-Apr-11

Publishing Venue

The IP.com Prior Art Database

Abstract

In an embodiment, a parallel imaging method using cyclic matrices for medical imaging systems, exploits fundamental properties of the linear algebra of parallel imaging. Accordingly, the method provides an improved SMASH parallel imaging technique that overcomes some of the drawbacks of existing SMASH techniques, including long reconstruction time and degraded image quality due to uncorrected aliasing. In particular, this invention provides a SMASH technique with fast reconstruction times that are comparable to GEM, AUTO-SMASH, VD-AUTOSMASH and GRAPPA, but with better approximations for the reconstruction algorithm, resulting in better image quality.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 16% of the total text.

Page 1 of 24

A PARALLEL IMAGING METHOD USING CYCLIC MATRICES FOR MEDICAL

    IMAGING SYSTEMS FIELD OF THE INVENTION
[0001] This invention relates generally, to parallel imaging, and more particularly to, a parallel imaging method using cyclic matrices for medical imaging systems.

BACKGROUND OF THE INVENTION
[0002] Parallel imaging techniques reduce MRI data acquisition time by using an array of multiple surface coils for receiving the signal. Acquisition time is reduced by increasing the step size between phase encoding lines or equivalently, by reducing the field of view (FOV). If the object extends outside the reduced field of view, however, aliasing (or wrapping) occurs in the phase encoding direction. Parallel imaging techniques remove the aliasing using the surface coil B1 fields (also called sensitivities) to find the unaliased spin distribution.
[0003] Conventionally, several parallel imaging techniques are in existence, including SENSitivity Encoding (SENSE) (1) and SiMultaneous Acquisition of Spatial Harmonics (SMASH) (2). SENSE and SMASH remove aliasing in the image and k- space domains, respectively. Several variations of the SMASH technique have also been developed, including Generalized Encoding Matrix (GEM) (3), AUTO-SMASH (4), Variable Density (VD) AUTO-SMASH (5) GeneRalized Autocalibrating Partially Parallel Acquisition (GRAPPA) (6), and Generalized SMASH (7).
[0004] Parallel Imaging with Localized Sensitivities (PILS) (8) is a simple method that snips away part of the aliasing from each image and pastes together the remnants. Another parallel imaging technique called Sensitivity Profiles from an Array of Coils for Encoding and Reconstruction In Parallel (SPACE RIP) (9) does not fall into either category of image space or k-space technique.
[0005] The following are the known methods of parallel imaging:

Page 2 of 24

[0006] SPACE RIP inverts the encoding matrix E to obtain the pseudoinverse 1

E - .

= . SPACE-RIP is much more computationally

intensive than other parallel imaging methods because (1) the matrix E is large (typically 512x256 for example) and computing 1

E - is time consuming, and (2) the multiplication

1

E S

 - is also time consuming because of the size of E. An advantage of SPACE-RIP is that, if the encoding matrix E is measured accurately, the reconstruction is very accurate
(i.e. very little uncorrected aliasing results) since few approximations are made. The cyclic matrix method provides image quality nearly as good as SPACE-RIP, with much faster reconstruction.
[0007] SMASH and variations including GEM, AUTO-SMASH, Variable Density AUTO-SMASH and GRAPPA approximate the -1

A matrix as block diagonal instead of

cyclical and with zeros outside the diagonal blocks. Reconstruction times are comparable to the cyclic matrix method for appropriate choice of parameters that give similar sized matrices, but since -1

A is not actually block diagonal, the image quality is somewhat

worse than for the cyclic matrix method (more...

Processing...
Loading...