Determination of appropriate energy threshold range for accurate estimation of effective atomic number considering statistical uncertainty in photon-counting techniques

Abstract

Purpose

The energy threshold is an important parameter for precise material identification employing photon-counting techniques. However, in such applications, the appropriate energy threshold has not been clarified. Therefore, we aimed to determine the appropriate energy threshold range for precise material identification, focusing on effective atomic number (Z) values as an index.

Methods

The atomic number was estimated using a previously proposed algorithm and Monte Carlo simulations. This algorithm included three steps: calculating the attenuation factor from the incident photon counts on a photon-counting detector, correcting the beam-hardening effects, and estimating the atomic number from the attenuation factor index using the calibration curve. Monte Carlo simulations were performed to add Poisson noise to an ideal x-ray spectrum. The total number of incident x-rays was set in the range of 10³–10⁶. The x-ray spectra were generated at tube voltages of 50–120 kV. Polymethyl methacrylate (Z = 6.5) and aluminum (Z = 13) were used for the analysis. The energy threshold was varied at intervals of 1 keV to estimate the atomic number. We evaluated the appropriate energy threshold range for accurately estimating the atomic number using the obtained atomic number data and statistical uncertainty under various conditions.

Results

The appropriate energy threshold range was found to be 31–38 keV for a tube voltage range of 50–120 kV. At this energy threshold, the atomic number can be estimated within an accuracy of ± 0.7 at 10⁵ counts for the atomic number range of 6.5 (PMMA) to 13 (Al).

Conclusions

We found the appropriate energy threshold range. The findings of this study are expected to be useful for appropriately setting the energy threshold during precise material identification using photon-counting detectors for clinical applications.