More on DQE
I commend Medical Imaging on taking on the difficult but important subject of quantitatively measuring image quality with digital detectors (“MTF, DQE and the Great Digital Detector Debate,” May 2001, p. 72). The insight, along with the comments of Dr. Dobbins from Duke University Medical Center and Dr. Maidment of Thomas Jefferson University Hospital, was very refreshing.

I want to expand on two of the points raised in the article, namely that DQE is not a point, but a curve and the speed/dose trade-off.

DQE is not a point but a curve. Most of the manufacturing literature and some of the scientific literature leads the reader to believe that measurements of image quality like Modulation Transfer Function (MTF) and Detective Quantum Efficiency (DQE) produce a single “score.” Even the article implies the same. In the figure titled “Comparison of Detector Image Quality” the text states that “Direct detectors generally have better MTF, but the a-Se direct detectors generally stop fewer X-rays than the indirect CsI:Tl detectors, thus they have lower DQE.” This begs the question of lower DQE, or better MTF, at what spatial frequency?

When comparing the performance of a system using MTF or DQE measures, it is important to consider the relevant portions of the DQE and MTF curves and not just a single spatial frequency such as 0 cycles/mm.

As Dr. Dobbins and Dr. Maidment point out, the finer the structure, the higher the resolution needed for faithful image representations. The diagnostic range for small bones is between 1 and 2.5 cycles/mm, while the diagnostic range for examining the intestine is between 0.3 and 0.6 cycles/mm. In general, the diagnostic range for general radiographic imaging is between 0.5 and 3.0 cycles/mm. This is the portion of the MTF or DQE curve where detector merit is most important and this is where the comparison between different systems should be made

While it is true that at zero spatial frequency the current versions of a-Se detectors have lower DQE than some CsI:Tl systems, the curves for these systems cross and the selenium detector exhibits both higher MTF and higher DQE for spatial frequencies in the areas of greatest interest to most radiologists.

Continuing in this same vein, there is no technical limitation prohibiting selenium detector manufacturers from depositing selenium in a sufficiently thick layer so as to have higher x-ray absorption than CsI:Tl and exhibit higher DQE even at zero spatial frequency. By comparison, making CsI:Tl thicker for indirect detection detectors is not useful, because the DQE would fall off rapidly at the higher spatial frequencies.

Speed/Dose Trade-Off. Detectors designed for digital mammography highlight the speed/dose performance tradeoff issue inherent in systems that use scintillators like CsI-based systems. These systems must have a sufficiently thick CsI layer so that the speed of the systems is not inferior to screen-film mammography. However, scintillator systems exhibit relatively poor resolution compared to screen-film. In comparison, selenium can be easily made essentially 100 percent absorbing at the low mammographic energies, and result in a digital detector having superior DQE (and MTF) over their entire spatial frequency range (from 0 to Nyquist) compared to either screen-film or CsI:Tl systems.

Andrew Smith, Ph.D.
Imaging Sciences
Hologic, Inc.