Algorithms for spectral calibration of energy-resolving small-pixel detectors

Abstract

Small pixel Cd(Zn)Te detectors often suffer frominter-pixel variations in gain, resulting in shifts in the individual energyspectra. These gain variations are mainly caused by inclusions and defectswithin the crystal structure, which affect the charge transport within thematerial causing a decrease in the signal pulse height. In imagingapplications, spectra are commonly integrated over a particular peak ofinterest. This means that the individual pixels must be accuratelycalibrated to ensure that the same portion of the spectrum is integrated inevery pixel. The development of large-area detectors with fine pixel pitchnecessitates automated algorithms for this spectral calibration, due to thevery large number of pixels. Algorithms for automatic spectral calibrationrequire accurate determination of characteristic x-ray or photopeakpositions on a pixelwise basis. In this study, we compare two peak searchingspectral calibration algorithms for a small-pixel CdTe detector in gammaspectroscopic imaging. The first algorithm uses rigid search ranges toidentify peaks in each pixel spectrum, based on the average peak positionsacross all pixels. The second algorithm scales the search ranges on thebasis of the position of the highest-energy peak relative to the averageacross all pixels. In test spectra acquired with Tc-99m, we found that therigid search algorithm failed to correctly identify the target calibratonpeaks in up to 4% of pixels. In contrast, the scaled search algorithmfailed in only 0.16% of pixels. Failures in the scaled search algorithmwere attributed to the presence of noise events above the main photopeak,and possible non-linearities in the spectral response in a small number ofpixels. We conclude that a peak searching algorithm based on scaling knownpeak spacings is simple to implement and performs well for the spectralcalibration of pixellated radiation detectors.