Contrast, noise, and blur affect performance and appreciation of digital radiographs

Abstract

We have studied the effect of the simultaneous variation of contrast, noise, and blur on the perceived quality of digital radiograph images. Three images were processed with different levels of roentgen photon noise, different luminance (gray-level) ranges, and different amounts of Gaussian blur. Observers had to rate the quality of the images on a numerical scale. To study how the quality judgement is related to the visibility of small details (diagnostic quality), and artificial detail was added to each image. The detail could be in two possible locations. The visibility of this detail was measured in a two-alternative forced-choice experiment. One of the conclusions is that the quality judgment was based mainly on blur, whereas the visibility of details was affected mostly by noise. © 1995 Society for Imaging Informatics in Medicine.