Development of active matrix flat panel imagers incorporating thin layers of polycrystalline HgI(2) for mammographic x-ray imaging

Abstract

Active matrix flat-panel imagers (AMFPIs) offer many advantages and have become ubiquitous across a wide variety of medical x-ray imaging applications. However, for mammography, the imaging performance of conventional AMFPIs incorporating CsI:Tl scintillators or a-Se photoconductors is limited by their relatively modest signal-to-noise ratio (SNR), particularly at low x-ray exposures or high spatial resolution. One strategy for overcoming this limitation involves the use of a high gain photoconductor such as mercuric iodide (HgI(2)) which has the potential to improve the SNR by virtue of its low effective work function (W(EFF)). In this study, the performance of direct-detection AMFPI prototypes employing relatively thin layers of polycrystalline HgI(2) operated under mammographic irradiation conditions over a range of 0.5 to 16.0 mR is presented. High x-ray sensitivity (corresponding to W(EFF) values of ∼19 eV), low dark current (<0.1 pA mm(-2)) and good spatial resolution, largely limited by the size of the pixel pitch, were observed. For one prototype, a detective quantum efficiency of ∼70% was observed at an x-ray exposure of ∼0.5 mR at 26 kVp.

Figure 1

Cascaded systems calculations of zero-frequency DQE plotted as a function of exposure for a variety of hypothetical direct and indirect detection active matrix flat-panel imager designs. These calculations assume a pixel pitch of 100 μm, a signal collection fill factor of 100% and an electronic additive noise level of 3000 e [rms]. The calculations further assume a 26 kVp x-ray spectrum with a Mo/Mo target/filter combination and use of a 5 cm BR-12 phantom. The dashed lines correspond to the performance of “conventional” AMFPI designs employing the type of direct and indirect detection x-ray converters commonly used for mammographic AMFPIs. These calculations assume a-Se and CsI:Tl converter thicknesses of 200 and 150 μm, densities of 4.3 and 4.51 g/cm³, and effective work functions (W_EFF) of 64 and 35 eV, respectively. In addition, calculations for AMFPI designs incorporating 100 μm thick HgI₂ converters with a density of 6.36 g/cm³ (corresponding to the single crystal form of the material) and W_EFF values of 5 and 20 eV, corresponding to the solid and dotted lines, respectively, are also shown.

Figure 2

(a) Dark current, normalized to unit photoconductor area, plotted as a function of photoconductor bias voltage. (b) Pixel x-ray signal plotted as a function of bias voltage. The vertical dashed lines indicate the operating voltage values selected for the remaining performance measurements.

Figure 3

Pixel x-ray signal plotted as a function of exposure. The lines correspond to fits to the data. The slopes of these lines, indicated in the figure, represent the x-ray sensitivity of each prototype.

Figure 4

(a), (b) Histograms of pixel x-ray signal at various exposures for prototypes 9B and 11B. The results have been corrected only for dark signal. (c), (d) Histograms corresponding to the same data as in (a) and (b), after the application of a gain correction obtained from flood field measurements (Antonuk et al., 1992) which nominally accounts for pixel-to-pixel variation in x-ray signal response. For purposes of comparison, dashed curves representing the behavior to be expected based solely on x-ray statistics, obtained through Monte-Carlo calculation of radiation transport, are superimposed on (c) and (d).

Figure 5

Measurements of MTF for prototypes 9B and 11B (open and closed symbols, respectively). The solid line is a sinc function based on a 127 μm aperture size. The MTF of a direct detection AMFPI with a 200 μm thick a-Se detector and the MTF for an indirect detection AMFPI with a 150 μm thick CsI:Tl detector (dashed and dashed-dot lines, respectively) are also shown. The plotted values for a-Se and CsI:Tl are based on published MTF results, (Zhao et al., 2003; El-Mohri et al., 2007) normalized to correspond to the 127 μm aperture of the MD88 array pixel pitch.

Figure 6

(a) DQE results for prototype 9B at exposures of 0.45, 1.92 and 2.71 mR. (b) DQE results for prototype 11B at exposures of 0.51, 1.3 and 1.9 mR. In addition, cascaded systems calculations for 127 μm pitch AMFPIs employing a 200 μm thick direct detection a-Se converter and a 150 μm thick indirect detection CsI:Tl converter are also shown. In each figure, the solid line corresponds to a theoretical prediction for the corresponding HgI₂ prototype obtained from cascaded systems analysis calculations. In addition, the dashed horizontal line corresponds to a theoretical upper limit on DQE(0) for the prototypes.

Figure 7

(a) and (b) Lag measured for prototypes 9B and 11B, respectively. Results are shown as a function of the dark frame number following an image frame. The data were acquired at a frame time of ~0.5 s and results are shown for three exposures.