Pulsed fluoroscopy detectability from interspersed adaptive forced-choice measurements.

Using a new reference/test, adaptive forced-choice method, we investigated detectability of low contrast disks in simulated x-ray fluoroscopy image sequences. The method minimized effects of undesirable variables, such as changes in subject attention level by alternating conventional fluoroscopy (reference) and low acquisition rate fluoroscopy (test) presentations. To compare absolute target detectability and x-ray dose requirements of reference and test presentations, the contrast of the test presentation was set equal to that of the reference presentation, and the dose of the test presentation was adapted to give equivalent detectability of disk targets. Comparing pulsed fluoroscopy at 15 acq/s (pulsed-15) and 7.5 acq/s (pulsed-7.5) with conventional fluoroscopy at 30 acq/s (pulsed-30), dose savings were found in 22 of 24 experiments. When averaged over 3 disk sizes and 4 subjects, savings were 22% and 26% for pulsed-15 and pulsed-7.5, respectively. In most experiments, the subject was given an unlimited time to view presentations. With unlimited display times, subjects took slightly longer to respond for pulsed-15 or pulsed-7.5 than for pulsed-30. In other experiments, display times were limited. As the restricted display time increased, absolute target detectability increased while the x-ray dose for equivalent detectability was unchanged. The number of frames, N, in a repeating loop was also varied. Absolute detectability increased as N increased from 1 to 30 but changed little as more frames were added up to 150. The new experimental paradigm gives efficient, robust comparisons of image data and is applicable to a wide variety of medical image perception problems.