Performance Trade-Off Analysis Comparing Different Front-End Configurations for a Digital X-ray Imager.

Performance of indirect digital x-ray imagers is typically limited by the front-end components. Present x-ray-to-light converting phosphors significantly reduce detector resolution due to stochastic blurring and k-fluorescent x-ray reabsorption. Thinner phosphors improve resolution at the cost of lowering quantum detection efficiency (QDE) and increasing Swank noise. Magnifying fiber optic tapers (FOTs) are commonly used to increase the field-of-view of small sensor imagers, such as CMOS, CCD, or electron-multiplying CCD (EMCCD) based detectors, which results in a reduction in detector sensitivity and further reduces the MTF. We investigate performance trade-offs for different front-end configurations coupled to an EMCCD sensor with 8 μm pixels. Six different columnar structured CsI(Tl) scintillators with thicknesses of 100, 200, 350, 500, and 1000 μm type high-light (HL) and a 350 μm type high-resolution (HR) (Hamamatsu) and four different FOTs with magnification ratios (M) of 1, 2.5, 3.3, and 4 were studied using the RQA5 x-ray spectrum. The relative signal of the different scintillators largely followed the relative QDE, indicating their light output per absorbed x-ray was similar, with the type HR CsI emitting 57% of the type HL. The efficiency of the FOTs was inversely proportional to M(2) with the M = 1 FOT transmitting 87% of the incident light. At 5 (10) cycles/mm, the CsI MTF was 0.38 (0.22), 0.33 (0.17), 0.37 (0.19), 0.23 (0.09), 0.19 (0.08), and 0.09 (0.03) for the 100, 200, 350HR, 350, 500, and 1000 μm CsI, respectively and the FOT MTF was 0.89 (0.84), 0.80 (0.72), 0.70 (0.60), and 0.69 (0.37) for M = 1, 2.5, 3.3, and 4, respectively. The 1000, 500, and 350HR μm CsI had the highest DQE for low, medium, and high spatial frequency ranges of 0 to 1.6, 1.6 to 4.5, and 4.5 to 10 cycles/mm, respectively. Larger FOT M resulted in a reduction in DQE. Quantifying performance of different front-end configurations will enable optimal selection of components for task-specific designs.