System performance of a prototype flat-panel imager operated under mammographic conditions.

The results of an empirical and theoretical investigation of the performance of a high-resolution, active matrix flat-panel imager performed under mammographic conditions are reported. The imager is based upon a prototype, indirect detection active matrix array incorporating a discrete photodiode in each pixel and a pixel-to-pixel pitch of 97 microm. The investigation involved three imager configurations corresponding to the use of three different x-ray converters with the array. The converters were a conventional Gd2O2S-based mammographic phosphor screen (Min-R) and two structured CsI:Tl scintillators: one optimized for high spatial resolution (FOS-HR) and the other for high light output (FOS-HL). Detective quantum efficiency for mammographic exposures ranging from approximately 2 to approximately 40 mR at 26 kVp were determined for each imager configuration through measurements of x-ray sensitivity, modulation transfer function (MTF), and noise power spectrum (NPS). All configurations were found to provide significant presampling MTF at frequencies beyond the Nyquist frequency of the array, approximately 5.2 mm(-1) , consistent with the high spatial resolution of the converters. In addition, the effect of additive electronic noise on the NPS was found to be significantly larger for the configuration with lower system gain (FOS-HR) than for the configurations with higher gain (Min-R, FOS-HL). The maximum DQE values obtained with the CsI:Tl scintillators were considerably greater than those obtained with the Min-R screen due to the significantly lower Swank noise of the scintillators. Moreover, DQE performance was found to degrade with decreasing exposure, although this exposure-dependence was considerably reduced for the higher gain configurations. Theoretical calculations based on the cascaded systems model were found to be in generally good agreement with these empirically determined NPS and DQE values. In this study, we provide an example of how cascaded systems modeling can be used to identify factors limiting system performance and to examine trade-offs between factors toward the goal of maximizing performance.