Multidimensional characterization of the perceptual quality of noise-reduced computed tomography images

Judging the perceptual quality of processed images is a cognitive process in which the perception of image attributes such as sharpness and noisiness plays an important role. In this paper, we use multidimensional scaling to study the perceptual factors that influence the quality impression of Computed Tomography (CT) images processed by a noise-reduction technique. We also characterize intersubject differences in the assessment of image attributes. We show that multidimensional scaling can be used reliably for the characterization of the subjective performance of image-processing algorithms. Evaluations using human subjects, such as the ones presented in this paper, will continue to play an important role, since objective measures for perceptual image quality, with proven validity in a broad range of applications, are not expected in the near future.