CdTe Based Energy Resolving, X-ray Photon Counting Detector Performance Assessment: The Effects of Charge Sharing Correction Algorithm Choice

Abstract

Most modern energy resolving, photon counting detectors employ small (sub 1 mm) pixels for high spatial resolution and low per pixel count rate requirements. These small pixels can suffer from a range of charge sharing effects (CSEs) that degrade both spectral analysis and imaging metrics. A range of charge sharing correction algorithms (CSCAs) have been proposed and validated by different groups to reduce CSEs, however their performance is often compared solely to the same system when no such corrections are made. In this paper, a combination of Monte Carlo and finite element methods are used to compare six different CSCAs with the case where no CSCA is employed, with respect to four different metrics: absolute detection efficiency, photopeak detection efficiency, relative coincidence counts, and binned spectral efficiency. The performance of the various CSCAs is explored when running on systems with pixel pitches ranging from 100 µm to 600µm, in 50 µm increments, and fluxes from 106 to 108 photons mm−2 s−1 are considered. Novel mechanistic explanations for the difference in performance of the various CSCAs are proposed and supported. This work represents a subset of a larger project in which pixel pitch, thickness, flux, and CSCA are all varied systematically.

Demonstrating a Novel, Hidden Source of Spectral Distortion in X-ray Photon Counting Detectors and Assessing Novel Trigger Schemes Proposed to Avoid It

“…The underlying simulations used the highest spectral resolution (130 energy bins), so that the spectral performance of each can be visualised the most clearly. The results shown are for illustrative purposes and assume a 1.5 mm thick sensor with a 250 µm pixel pitch, as these values were determined by us to be optimal for imaging a monoenergetic source at this energy in previous work [17,20]. The use of both raw and normalised energy spectra is helpful as each emphasises different aspects of the various triggering schemes: raw plots are better for relaying the counting efficiency of each scheme (as the areas under the curve represent the total number of counts) whilst plots normalised to maximum peak height are better for assessing spectral efficiency (by allowing the relative heights of each bin to be more easily compared).…”

“…80 keV was chosen as it is high enough to allow the expected peaks to be well separated whilst still occurring at medically relevant energies. This energy is also relevant to other work ongoing in our group, including the previous simulation work we have done with CoGI to optimise other x-CSI system parameters [17,20]. Sensor thickness 1 to 3 (steps of 0.5) mm…”

“…It should be remembered that the pixel pitch and sensor thickness used in generating Figure 19 were specifically chosen to maximise the performance of the STD scheme under the monoenergetic irradiation performed here, based on previous work [17,20]. For that reason, it is perhaps not surprising that STD was only clearly inferior to all other schemes at the highest fluxes investigated and may seem obvious that changes to the pitch or thickness could shift the point at which STD becomes the worst scheme.…”

Ballistic Deficit Pulse Processing in Cadmium–Zinc–Telluride Pixel Detectors for High-Flux X-ray Measurements

“…As it is well known, the energy of the shared events can be recovered by summing their energies, measured in temporal coincidence. However, the energy recovered after charge sharing addition (CSA) is often characterized by deficits, due to the presence of charge losses at the inter-pixel gap [57][58][59][60][61][62][63][64][65]. In our case, we investigated the effects of ballistic deficit pulse processing on these charge losses.…”