Contrast and Noise

Contrast and noise are measured from the aluminum square in the middle of the phantom and an adjacent background region of interest to produce Contrast to Noise (CNR) and Signal to Noise (SNR). Equal sized areas are assessed in the contrast object and background to calculate the CNR and SNR. Contrast is calculated by the difference in median CT numbers, or median pixel values, between the contrast object ROI and the background ROI. The median is used instead of the mean, in case there’s an artifact that skews the mean pixel value. The noise is estimated as the standard deviation of CT numbers in the background ROI. Below is an annotated picture and the equations used to calculate CNR and SNR.

Figure 2.1: Zoomed in CT image of the 0.2mm thick, 1cm aluminum foil square test object surrounded by the 8 low contrast (acrylic) spheres. The aluminum square can be used to calculate the CNR, or its edge can be used to calculate the line spread function (LSF) and corresponding modulation transfer function (MTF) in the x and y directions. 

CNR = (Median Pixel Value of Contrast Object - Median Value of Background ROI)/ (Standard Deviation of Background ROI)

SNR = (Median Value of Background ROI)/ (Standard Deviation of Background ROI)