Hongye Liang, Abstract Co-Author: Nothing to Disclose

Aldo Badano PhD, Presenter: Nothing to Disclose

Brandon D. Gallas, Abstract Co-Author: Nothing to Disclose

Subok Park PhD, Abstract Co-Author: Nothing to Disclose

Kyle Jean Myers PhD, Abstract Co-Author: Nothing to Disclose

To quantify the effect of slow display devices on lesion detectability in stack-mode presentation of large image sets in medical liquid crystal displays (LCDs).

Measurements of all possible (256x255) gray-level transition times were performed on medical and non-medical LCDs. The temporal response matrix is then transformed into a luminance matrix where each element represents the luminance achieved after one frame-time. The measurement results are validated with fast acquisitions of luminance fields from the screen at more than 100 fps. We then use a contrast-sensitive channelized-Hotelling computational observer modified for temporal reading to determine the effect of slow display on detectability with 3D clustered lumpy backgrounds and a 3D designer nodule as the signal. The model observer, shown to predict human performance for non-Gaussian backgrounds, is validated with human reader studies performed on a CRT (with all rise and fall times below 8 ms).

Our calculations of contrast degradation show that for specific transitions for which the display temporal response is poor, the luminance after 1, 3, and 5 frames reaches only 12, 45, and 75% of the target value, respectively. If the number of images in the stack approaches one per frame, the reduction in contrast is severe with peak luminance as low as 50% of the target value. At a browsing speed of 30 frames/second, the average luminance contrast goes down to 30% of the original value. Our study suggests that the magnitude of the effect depends on gray level of the background and signal. Our human and model observer calculations suggest that reduction in detection performance parallels the reduction in luminance contrast concomitantly.

We find that slow response of the display device greatly affects lesion contrast and observer performance. Poor temporal response of display devices used for stack-mode reading of large image sets can degrade the performance of the radiologist by reducing the effective luminance contrast.

This methodology, after further validation with human observers, could be used to set limits for the rendering speed of large image sets (from CT, MR, or tomosynthesis) when read in stack mode.

Liang, H, Badano, A, Gallas, B, Park, S, Myers, K, Detectability Decreases When Browsing Large Image Sets in Slow Display Devices.  Radiological Society of North America 2007 Scientific Assembly and Annual Meeting, November 25 - November 30, 2007 ,Chicago IL. http://archive.rsna.org/2007/5009577.html