Abstract

The goal of this study was to improve detectability of low contrast features within moving coronary arteries in dynamically displayed cineangiograms. We compared performance and decision times between the standard display, in which the artery moves with respect to a nearly stationary background of patient structure noise, and a "stabilized" display, where each image of the sequence is digitally shifted so that the artery is fixed at the center of the screen and the background moves. We performed a four-alternative forced-choice study to measure detectability of spherical voids within cylindrical segments (simulating arteries) mathematically projected into human coronary angiograms to produce images with realistic quantum and structure noise. For both display types the resulting 32-frame cine loops were displayed at 7.5, 15, and 32 frames per second. Four subjects made 320 observations for each display type at each frame rate. There was no time limit, but time for each response was recorded. Results show that the stabilized display at 32 fps improved detectability (d′) over the best standard display (moving artery at 8 fps) by an average of 50%. Furthermore, the optimized display reduced the average decision time by a factor of 3. We conclude that this display may be of significant value in coronary angiography and in similar visual detection tasks in dynamic displays.

© 1992 Optical Society of America