Sensors 2020, 20, 6093 5 of 24
trilinear interpolation of the relevant CIE maps from the finite element method, resulting in a list
of pre-thresholded signals. SGS then used the rules of a given CSCA (or the no CSCA rules) to run
through the list and identify groups of events that need to be summed into a single readout event.
The total signal produced by these events is then compared to the preset energy thresholds so that the
event can be assigned to the correct energy bin. In a physical system all counters that the event passes
would be incremented by one and then the number of events in each bin determined by subtraction,
however due to the large amount of data being processed in this work the more computationally
expedient approach of assigning signals directly to their correct energy bin was adopted. The resulting
data set contains for each pixel a list of output signal intensities and times, which can be used to
construct an image or aggregated to assess the performance of the detector as a whole. Due to the flat
field irradiation used, the aggregated readout scheme was used in this work.
2.4. Charge Sharing Correction Algorithms Considered
All CSCAs considered in this work function in a similar way, by identifying events that occur in
nearby pixels within some time window and summing them together to form a single event. The theory
behind this approach is that events that are spatially and temporally close together are more likely to
result from a single photon generating charge across multiple pixels than to originate from multiple
independent photons. By summing these events together then it is hoped that the original energy of
the incident photon can be reconstructed. The CSCAs considered in this work were classified with
respect to two terms: neighbourhood size and neighbourhood locality.
Neighbourhood size (NS) refers to the size of the search area, in pixels, over which events were
identified for summing. 2
×
2 CSCAs consisted of a 2-pixel by 2-pixel search area, whilst 3
×
3 CSCAs
utilized a 3-pixel by 3-pixel search area.
Neighbourhood locality (NL) refers to the way in which the search area (neighbourhood) is defined,
as shown in Figure 1. Static CSCAs use a predefined scheme to link all pixels to a unique NxN
neighbourhood of pixels that does not change throughout the irradiation. In contrast, dynamic CSCAs
define the NxN neighbourhood relevant to each event only when that event is detected, using a predefined
rule. For the dynamic 2
×
2 CSCA in this work that rule is that the event is placed in the bottom left pixel
of a 2
×
2 search area. For the dynamic 3
×
3 CSCA in this work the rule is that the event is placed in the
middle pixel of a 3
×
3 search area. Due to this on-the-fly neighbourhood construction, dynamic CSCAs
may lead to a case where proposed neighbourhoods would overlap. Pixels called multiple times in quick
succession like this report their full charge to the first neighbourhood they are assigned to, but a charge
of zero to all subsequent calls, until they have reset and are ready to detect another event.
Sensors 2020, 20, x FOR PEER REVIEW 5 of 24
The total signal produced by these events is then compared to the preset energy thresholds so that
the event can be assigned to the correct energy bin. In a physical system all counters that the event
passes would be incremented by one and then the number of events in each bin determined by
subtraction, however due to the large amount of data being processed in this work the more
computationally expedient approach of assigning signals directly to their correct energy bin was
adopted. The resulting data set contains for each pixel a list of output signal intensities and times,
which can be used to construct an image or aggregated to assess the performance of the detector as a
whole. Due to the flat field irradiation used, the aggregated readout scheme was used in this work.
2.4. Charge Sharing Correction Algorithms Considered
All CSCAs considered in this work function in a similar way, by identifying events that occur in
nearby pixels within some time window and summing them together to form a single event. The
theory behind this approach is that events that are spatially and temporally close together are more
likely to result from a single photon generating charge across multiple pixels than to originate from
multiple independent photons. By summing these events together then it is hoped that the original
energy of the incident photon can be reconstructed. The CSCAs considered in this work were
classified with respect to two terms: neighbourhood size and neighbourhood locality.
Neighbourhood size (NS) refers to the size of the search area, in pixels, over which events were
identified for summing. 2 × 2 CSCAs consisted of a 2-pixel by 2-pixel search area, whilst 3 × 3 CSCAs
utilized a 3-pixel by 3-pixel search area.
Neighbourhood locality (NL) refers to the way in which the search area (neighbourhood) is
defined, as shown in Figure 1. Static CSCAs use a predefined scheme to link all pixels to a unique
NxN neighbourhood of pixels that does not change throughout the irradiation. In contrast, dynamic
CSCAs define the NxN neighbourhood relevant to each event only when that event is detected, using
a predefined rule. For the dynamic 2 × 2 CSCA in this work that rule is that the event is placed in the
bottom left pixel of a 2 × 2 search area. For the dynamic 3 × 3 CSCA in this work the rule is that the
event is placed in the middle pixel of a 3 × 3 search area. Due to this on-the-fly neighbourhood
construction, dynamic CSCAs may lead to a case where proposed neighbourhoods would overlap.
Pixels called multiple times in quick succession like this report their full charge to the first
neighbourhood they are assigned to, but a charge of zero to all subsequent calls, until they have reset
and are ready to detect another event.
(a) (b)
Figure 1. A 16 pixel array containing 3 detected events, with the red star representing the first of these
events to be detected. (a) Static charge sharing correction algorithms (CSCAs) predefine the various
search areas (shades of grey) before any events are detected. (b) In contrast, dynamic CSCAs define
search areas after an event is detected based on a predetermined rule. In this case, the dynamic 2 × 2
CSCA is being applied, so the 2 × 2 search area is defined such that the first detected event is in the
bottom left pixel of the search area.
Figure 1.
A 16 pixel array containing 3 detected events, with the red star representing the first of these
events to be detected. (
a
) Static charge sharing correction algorithms (CSCAs) predefine the various
search areas (shades of grey) before any events are detected. (
b
) In contrast, dynamic CSCAs define
search areas after an event is detected based on a predetermined rule. In this case, the dynamic 2
×
2
CSCA is being applied, so the 2
×
2 search area is defined such that the first detected event is in the
bottom left pixel of the search area.