An Approach of Deriving Relative Sensitivity Profiles for Image Reconstruction in MRI

The determination of accurate coil sensitivity profiles is crucial for parallel MRI approaches such as SMASH, SENSE, and SPACE-RIP. In classical SENSE the profiles are determined using a so-called calibration scan through division of array and body coil images sampled with full field of view (FOV). However, as this relies on the availability of a body coil, there exist many situations where this is not possible. This research work outlines the method of calculating sensitivity maps of array coil elements relative to one particular reference array element using complex weighting coefficients, based on a few lines of central
k
-space data, similar to those used in the GRAPPA method. The approach provides relative sensitivities with a smooth boundary between objects and image background. When the relative sensitivities are used in a SENSE algorithm, referred to as relative SENSE (rSENSE), the reconstructed image corresponds to a fully sampled image taken by the reference array element. The images of other elements are then obtained by multiplying the reference image by their relative sensitivities. Results provided for the proposed approach of deriving relative sensitivity profiles for fast image reconstruction show that the technique provides images with fewer artifacts. The findings are established with reference to SFDD, SPDD, and GRAPPA.