KAHIBARO
19.1. What Are Coincidences?
Table of Contents
Coincidence detection
In PET simulations and other nuclear imaging setups, a coincidence represents two or more detector events that are related to each other and are treated as a single measurement. In most GATE PET applications, a coincidence is a pair of detector “singles” that are linked because they arise from the same physical emission process.
In PET, a positron emitted from a radionuclide such as F‑18 travels a short distance, slows down, and annihilates with an electron. This annihilation typically produces two 511 keV photons that fly out in almost opposite directions. If these two photons are detected in two different crystals of the scanner within a short time, the system considers these detections as a coincidence event. The line that joins the two detection positions is called a line of response and is one of the basic inputs for PET image reconstruction. The underlying idea is that the annihilation took place somewhere along this line.
In GATE, individual detector interactions go through a digitizer chain where hits are processed into singles. A coincidence module or actor then analyzes these singles and forms coincidence events according to rules that you define. These rules include which singles can be paired, the time interval allowed between them, and often additional conditions such as energy selection. The output is a list of coincidence events, where each entry usually contains information about the two involved detectors, positions, energies, detection times, and identifiers that help to classify the event.
Coincidence detection is not limited to PET. Any system that uses multiple detectors to identify correlated particles can use similar logic. However, in this course, coincidences mainly refer to PET-style timing correlations between gamma detections. Detailed classification into true, scattered, and random coincidences is addressed in later sections and chapters, so here the focus is on the basic concept that a coincidence is a grouped set of singles that are assumed to belong to the same physical emission.
It is important to understand that coincidence detection is a software or electronics decision built on top of individual detections. Two photons may arrive at the scanner very close in time, but they are only turned into a coincidence event when the sorting algorithm, either in GATE or in real hardware, decides that they satisfy the required conditions.
Timing windows
The key tool used to decide whether two singles belong to the same coincidence is the coincidence timing window. This is a predefined time interval. If the difference between the detection times of two singles is smaller than this interval, the system considers them as a coincidence; if it is larger, they are treated as unrelated events.
In a simple PET setup, each single has a timestamp, for example $t_1$ and $t_2$. The time difference is
$$
\Delta t = |t_1 - t_2|.
$$
The coincidence condition can be written as
$$
\Delta t \leq \Delta t_{\text{window}},
$$
where $\Delta t_{\text{window}}$ is the width of the coincidence time window that you configure.
The basic coincidence rule is:
$$
\Delta t = |t_1 - t_2| \leq \Delta t_{\text{window}}.
$$
If this condition is not satisfied, the two singles do not form a coincidence.
A narrow timing window reduces the probability that two independent, unrelated photons will accidentally fall into the same time interval and be counted as a coincidence. This lowers the random coincidence rate and usually improves the quality of the data. However, if the window is too narrow, true coincidence events where both annihilation photons were genuinely emitted together may fall outside the window because of detector and electronics timing uncertainties. These true events are then lost, which decreases sensitivity.
A wider timing window does the opposite. It increases the number of accepted true coincidences, but it also increases the number of random coincidences. In practice, the window is chosen to balance sensitivity and randoms. The optimal value depends on the detector timing resolution, the activity in the field of view, the scanner design, and the application.
In GATE, the timing window is part of the coincidence sorting configuration in the digitizer chain. You typically specify a coincidence time window in units such as nanoseconds. The digitizer then scans the list of singles, groups those that fall within this window, and writes the resulting coincidence events to an output file. More advanced time handling, such as using time-of-flight information, still relies on the same core idea: the difference between timestamps decides which singles are grouped and also where along the line of response the annihilation is localized.
Views: 10