[MRI image reconstruction using polar-coordinates conversion of k-space data].

In this study, we proposed the new reconstruction techniques for magnetic resonance imaging (MRI) using filtered back projection (FBP) or simultaneous reconstruction technique (SIRT). We converted the k-space which was acquired by conventional phase-encoding schemes from Cartesian coordinates to polar coordinates and created the projection. The linear interpolation and the sinc interpolation were used in the conversion. The accuracy of the reconstructed image using projection was evaluated by the relative error in comparison with the standard image which was reconstructed by the two-dimensional Fourier transform (2DFT) with conventional Cartesian k-space. The relative error reconstructed both FBP and SIRT from projection with sinc interpolation is 0.013. The maximum value of standard image is 1.501451, FBP is 1.47921, and SIRT with iteration 100 is 1.44858 and with iteration 200 is 1.579442. The minimum value of both the standard image and the others is about 0. Visually, there is no margin between the standard image and the reconstructed image from projection with FBP or SIRT.