Computation of the modulation transfer function for an MR imaging system using a polysaccharide tissue-mimicking phantom

A comprehensive experimental study was performed on the application of the Modulation Transfer Function (MTF) principle on MR imaging system. MTF computed from an Edge Spread Function (ESF) was the preferred method for this work. Three conditions have to be established for a successful implementation of the MTF analysis technique: maintaining ESF signal-to-noise ratio equals 190 or more, registering correctly the ESF profile within the sampling window and preserving phases of all ringing amplitudes in the ESF profile.;The computed MTF was determined and evaluated for the spin echo imaging technique as well as for the fast imaging techniques Gradient Recalled Acquisition in the Steady State (GRASS), RF Spoiled Gradient Recalled (SPGR), and Fast Spin Echo (FSE).;Using a water/Plexiglas phantom and CuSO4/Plexiglass phantoms, factors like receiver bandwidth, location with the head coil, and T2 relaxation time were experimentally studied and their effects on the MR imaging system spatial frequency response were determined. Doubling the receiver bandwidth leads to a 30% loss in the MTF at 0.9 level. Main field inhomogeneity has some effects on the MR imaging response at different locations within the head coil. A CuSO4/Plexiglass T2 relaxation time around 55 msec showed some distortion in the shape of the GRASS MTF .;A Polysaccharide tissue mimicking MTF phantom was developed for the purpose of studying the MR system spatial frequency response under simulated clinical conditions. Loss of phase polarity by the magnitude operator was avoided by biasing the background of the ESF such that all ringing amplitudes have a positive polarity. Lowering the Polysaccharide concentration from 6% to 5% by weight did this background biasing shift. Computed MTF from the CuSO4/Polysaccharide tissue mimicking phantom showed that the MR imaging spatial frequency limit is strongly noise dependent.;For SE and FSE imaging techniques, the MTF computed from complex image data is the best method for characterizing the MR imaging spatial frequency response. Because imaging techniques based on the gradient recall principle, like GRASS and SPGR, are unable to correct for the main field inhomogeneity, the best method is the MTF computed from the ESF with a biased background level.