Application of a Fourier Shift Preprocessing Stage to Improve the Resolution of Resting State fMRI Images
Multiple sclerosis (MS) is a disease common in many northern-climate countries with Canada having 28% higher MS numbers on a population basis than second place Denmark. Optical Neuritis (ON) is known to affect the properties of the visual pathways in the brain, is often a precursor to MS, and has been suggested as a system model for MS pathology. We have investigated possible resting state functional magnetic resonance imaging (rs-fMRI) markers to track ON recovery or progression to MS. To obtain the necessary rs-fMRI temporal resolution requires discrete Fourier transform (DFT) reconstruction applied to 2D truncated (finite length) frequency domain MRI data sets followed by a 2D DFT-based correlation analysis across a time sequence of images to identify image regions that are connected through the brain's optical pathways. Another DFT-based transfer function determination identifies pathways impacted by ON; permitting differentiation between normal volunteers and ON patients. Windowing or low-pass filtering is required to remove ringing distortions from these five DFT application stages, but leads to lower fMRI spatial resolution and an undesirable loss in ON marker accuracy. Recently we have theoretically identified a Fourier shift manipulation (FSM) preprocessing stage that avoids the unnecessary loss of resolution that occurs with the use of global windowing during DFT application. We have previously demonstrated how applying FSM to data improves 1D DFT-based analysis under certain experimental MR-relevant situations. In this paper we extend the FSM approach to demonstrate an improvement in the 2D resolution of rs-fMRI images generated from truncated MRI k-space data.