ABSTRACT
Although it is common practice to digitize radionuclide images onto the finest matrix available, their low count densities and poor spatial resolution suggest that quite large pixels should be adequate. Observers find these large pixels visually obtrusive, but minification of the image can reduce this effect. Experiments reported here have investigated how minification (achieved by increasing viewing distance) affects the perceived quality of images digitized onto different sized matrices. Observers' subjective preference for different pixel sizes was measured at various viewing distances using clinical bone images as test patterns. An objective measure of image quality was made by comparing the detectability of computer-generated focal areas of increased activity both in simple noisy backgrounds and in clinical bone images. The results show that a 128 X 128 matrix is adequate when the image is 8 cm2 and is viewed from 1 and 2 m. A finer matrix failed to produce better results.
Subject(s)
Radionuclide Imaging/methods , Bone and Bones/diagnostic imaging , Evaluation Studies as TopicABSTRACT
The effect of picture element (pixel) size on image quality is investigated. Asked simply to state their preference for clinical images displayed with different pixel sizes, observers preferred those with the smallest pixel. Specific perceptual tasks, however, such as abnormality detection, are found to be insensitive to changes in pixel size up to a limit. Interpolation is found to have no effect on perceptual tasks, but 'random interpolation' produced a more pleasing image than simple linear interpolation.
