Improving digital charge sharing compensation in photon counting detectors with a low‐threshold comparator

• Published in: Medical physics (Lancaster) Vol. 48; no. 10; pp. 5819 - 5829
• Main Authors: Hsieh, Scott Sigao, Iniewski, Kris
• Format: Journal Article
• Language: English
• Published: 01-10-2021
• Subjects: charge sharing; dual energy; photon counting
• ISSN: 0094-2405; 2473-4209
• DOI: 10.1002/mp.15154

Purpose
Charge sharing is a major non‐ideality in photon counting detectors (PCDs) and can increase variance in material decomposition images. Analog charge summing (ACS) is an effective mechanism for charge sharing compensation (CSC), but is complex to implement and may limit the maximum count rate of the PCD. Digital CSC mechanisms such as digital count summing (DCS) may be simpler to implement; however, earlier simulation studies suggest that digital CSC only provides half the benefit of ACS. We propose including an additional low‐threshold comparator (LTC) underneath the noise floor of the PCD to improve the effectiveness of digital CSC.
Methods
We simulated a PCD with four or eight equally spaced energy bins. X‐ray photons arrived on the PCD following a Poisson distribution, and charge was allocated to PCD pixels following Monte Carlo techniques. Gaussian electronic noise was added with standard deviation of 2 keV and the signals were processed with four CSC schemes: no CSC, ACS, DCS, and DCS with LTC. The energy bins were placed from 25 to 100 keV at 25 keV intervals (for four bins) or from 25 to 112.5 keV at 12.5 keV intervals (for eight bins), and the LTC threshold was placed at 8 keV in both cases. The binned counts were transformed into estimates of water and iodine material thickness using a linear estimator that was fitted to the data. Our simulations were performed in the low‐flux limit without any pileup, assuming a 120 kVp spectrum, 25 cm water object, and 0.3 mm PCD pixel size.
Results
All CSC schemes decreased variance in basis material decomposition. In the four‐bin PCD, the relative dose efficiencies (inverse of the variance) for iodine material decomposition were 1.0, 2.4, 3.2, and 4.3 for a PCD without CSC, DCS without LTC, DCS with LTC, and ACS, respectively. In the eight‐bin PCD, the relative dose efficiencies were 1.1, 2.5, 3.1, and 4.8, respectively. In a sensitivity analysis, electronic noise had a stronger deleterious effect on ACS than DCS. In simulated visual images, DCS and ACS improved high frequency contrast in material decomposition images.
Conclusions
Introducing an LTC may reduce the performance differential between DCS and ACS. These findings have been derived from simulation studies only and have not been validated experimentally.