Computational phantoms for investigating impact of noise magnitude on modulation transfer function

Abstract

Accurate measurement of spatial resolution in terms of modulation transfer function (MTF) is essential in computed tomography (CT) images. The purpose of this study was to developed a computational phantom that can be used to evaluate the effect of noise on the MTF in CT images. Our computational phantoms for measuring MTF in CT were developed with MATLAB software. The phantom image was blurred by a point spread function of a certain standard deviation. Subsequently, different noise levels were added to the phantoms. Next, an automatic MTF calculation was implemented. The first step of the MTF calculation was to determine the region of interest (ROI). Profile was generated from the ROI, and a line spread function (LSF) curve was formed. The LSF curve was Fourier transformed to produce a MTF curve. Greater noise added to phantom image, it yields greater effect of standard deviation on the measured MTF. The greater noise makes the MTF curve increases differently than MTF with 0 HU noise. The 10% MTF values at the 25% noise reach more than 2.0 cycle/mm. By the developed computational phantoms, the spatial resolution and the amount of noise can be determined independently.