The Effects of Spectral X-Ray Photon Counting Detector Parameters on Detector Performance: Thickness and Pitch

Oliver Len Paul Pickford Scienti,Dimitra G Darambara,Jeffrey C Bamber

doi:10.1109/access.2020.3033969

Oliver Len Paul Pickford Scienti, Dimitra G Darambara + Show 1 more

Open Access

https://doi.org/10.1109/access.2020.3033969

Copy DOI

Abstract

The design of photon counting spectral imaging (x-CSI) X-ray detectors involves optimising many parameters including pixel thickness, pitch, and type of charge sharing correction algorithm (CSCA) employed, if any. The optimal value for one parameter depends on the other parameter values, as well as extrinsic variables such as application specific photon fluxes and energies of interest. No analytical approaches currently exist for optimising these parameters simultaneously. This work thus utilised our inhouse simulation framework, combining Monte Carlo and finite element methods to systematically simulate the response of 715 different CdTe based x-CSI detectors, comprising 13 different CSCAs, 5 different pixel thicknesses (1 mm - 3 mm), and 11 different pixel pitches (100 μm - 600 μm). Detector response to monoenergetic irradiation at 80 keV at 4 different fluxes was assessed using a range of metrics. Due to its complexity, the analysis of this work is divided into several publications, with this one focusing on the effects of pixel pitch and thickness. We were able to identify, and provide mechanistic explanations for, general trends in detector performance with varying pixel geometry that will be of interest to x-CSI detectors designers. Superficially similar spectral metrics were found to vary significantly in their sensitivity to different charge sharing mechanisms, underlining the importance of carefully selecting the evaluation metric for photon counting detectors based on their application. The parameters used here were selected based on our own interests, however this work demonstrates the utility of this framework for optimising x-CSI detector parameters for various spectral applications.

Highlights

Traditional x-ray imaging technologies make use of energy integrating (EI) approaches, whereby the signal output from a given pixel is the integral of charge in the circuit over some time interval, t
As discussed previously, the results shown concern a single flux (~107 photons s-1 mm-2) and are raw data readouts from the sensors, with no charge sharing correction algorithms applied, unless otherwise stated in the figure captions
It is of note that neither of these trends is linear in nature and that for many of the geometries shown the Absolute Detection Efficiency (ADE) is greater than 100%

Summary

Introduction

Traditional x-ray imaging technologies make use of energy integrating (EI) approaches, whereby the signal output from a given pixel is the integral of charge in the circuit over some time interval, t. In order to minimise the contribution of the electronic noise to the output signal, the time interval t is set to be long enough that many photons will deposit their energy in the pixel during the process of charge integration, and so the fractional charge due to electronic noise is minimised. This approach has three related drawbacks: 1). These detectors still use spectra with significant spectral overlap as standard [1], so involve a higher radiation dose than would be needed if the energy dependent attenuation could be assessed directly

Methods

Results

Discussion

Conclusion