Improvement of Image Reconstruction Techniques for μMRI

Abstract

In microscopic magnetic resonance imaging (μMRI), time-domain signal is sampled on a nuclear magnetic resonance (NMR) spectrometer, then transformed into the frequency domain as the image. The conventional technique of data sampling traverses on a uniform Cartesian grid, which can be reconstructed into images by the application of the Fast Fourier Transform (FFT). In specific cases, the data sampling uses a radial grid, which needs to be interpolated onto the standard Cartesian grid before transformation. Existing methods are competitive but could be improved. This paper explores alternative image reconstruction algorithms for the radially sampled time-domain data to minimize the image artifacts due to the reconstruction of the Cartesian grid.