KAHIBARO
17.7. Energy Thresholds
Table of Contents
Lower threshold
Every real detector ignores very small signals. In GATE, this behavior is modeled with an energy threshold applied during digitization, typically at the singles level. The lower threshold represents the minimum detected energy that is considered a valid signal in a detector channel. Any hit or summed energy in a channel that falls below this limit is discarded and does not appear in the singles output.
When simulated particles deposit energy in a crystal or pixel, a digitizer usually first sums all hit energies that belong to one detector channel within a given time window, then applies blurring to account for detector energy resolution, and then compares the blurred energy to the defined thresholds. For the lower threshold, the key rule is simple.
Any event with detected energy $E < E_{\text{low}}$ is rejected and removed from the digitized output.
The lower threshold serves multiple purposes. It suppresses electronic noise that would otherwise produce low-amplitude signals, removes tiny partial energy deposits that are not useful for imaging, and can reduce the volume of stored data by eliminating events that carry little or no useful information. In nuclear medicine detectors and PET scanners, this threshold is often chosen to be well below the photopeak to keep most scattered and photopeak events, for example around 350 keV for 511 keV photons. In SPECT and gamma cameras, it is usually set below the main photopeak of the radionuclide, for example below the 140 keV photopeak of Tc 99m.
The chosen $E_{\text{low}}$ has a direct influence on sensitivity and scatter fraction. If the lower threshold is too low, many scattered and noisy events will survive, which increases background and can degrade image contrast. If it is too high, a significant number of good photopeak events are lost, which reduces sensitivity and increases statistical noise. In practice, you will often start with a lower threshold that matches clinical practice for the modality you are simulating, then adjust it based on your specific research goals, for example higher to study the impact of scatter rejection, or lower to increase statistics for qualitative tests.
In GATE, thresholds are configured as parameters of the digitizer chain. You typically specify a unit aware value, such as 350 * keV, and attach it at the appropriate stage, usually to the module that creates singles from hits. Because the same digitizer settings apply to all relevant detector channels in that chain, the lower threshold is usually uniform across the system, unless you explicitly define separate digitizers for different detector regions.
Upper threshold
The upper threshold is the counterpart of the lower threshold and represents the maximum detected energy that you accept as a valid event. In many clinical imaging systems, very high energy deposits are rare or uninteresting for the main photopeak. They may be associated with pile up, wrong calibration, or interactions of particles that are not part of the imaging process you want to study. The upper threshold allows you to remove these outliers.
The principle is again straightforward.
Any event with detected energy $E > E_{\text{high}}$ is rejected and removed from the digitized output.
For PET, the upper threshold is commonly chosen slightly above the 511 keV photopeak, which bounds the main energy window. For example, if you use an energy window from 350 keV to 650 keV, then 650 keV effectively plays the role of an upper threshold. For SPECT, the upper threshold is usually above the main photopeak, such as 154 keV for Tc 99m when the photopeak is at 140 keV.
The impact of the upper threshold is somewhat different from the lower one. If it is too low, some valid full energy events will be cut off, especially in the presence of realistic energy blurring, where true photopeak events can appear at slightly higher energies. This leads to a loss of sensitivity and possible distortions in the measured energy spectrum. If the upper threshold is too high, then events with higher energies, including some from unwanted background or other radionuclides, may be retained. This can increase contamination from non primary photons or misidentified events.
In research simulations, the upper threshold is also used to define scatter or auxiliary energy windows. For example, a SPECT simulation might define a scatter window just below the photopeak, and treat energies above a certain level differently. In such cases, you often configure several windows, each with its own lower and upper limits, instead of a single global upper cut.
In the GATE digitizer chain, upper thresholds are specified with units in the same way as lower thresholds. They act at the level of processed detector signals, so any earlier details such as the number of hits or local energy fluctuations are already compressed into a single energy value per channel. When you debug your simulation, it is good practice to inspect the energy spectrum of singles before and after applying thresholds, to confirm that the chosen $E_{\text{high}}$ does not cut into the main photopeak region.
Energy windows
Energy thresholds become especially powerful when you combine them to create energy windows. An energy window is defined by a pair of limits, a lower and an upper threshold, and selects only those events whose detected energy lies within that interval. This is how most gamma imaging systems isolate the photopeak region or define scatter windows.
Mathematically, an energy window from $E_{\text{low}}$ to $E_{\text{high}}$ accepts events that satisfy the condition
An event belongs to an energy window if and only if
$$
E_{\text{low}} \leq E \leq E_{\text{high}}.
$$
Energy windows are usually chosen around a known photopeak energy $E_0$, such as 511 keV for PET or 140 keV for Tc 99m SPECT. A typical way to specify them is by a percentage width around $E_0$. For a symmetric window with a relative half width $w$ (for example 10 percent), the limits are
$$
E_{\text{low}} = E_0 \, (1 - w), \quad E_{\text{high}} = E_0 \, (1 + w).
$$
In practice, windows can also be asymmetric to better match detector response or to exclude specific interference regions. For example, a gamma camera might use a primary photopeak window centered at the radionuclide energy and one or more adjacent scatter windows that capture scattered photons at lower energy. These additional windows are very useful for scatter correction studies, where you want to estimate the scatter contribution separately.
The role of an energy window is to balance two competing objectives. If the window is wide, you retain more counts, which improves statistical precision but includes more scattered and misidentified events, so image contrast or quantitative accuracy can degrade. If the window is narrow, you reject more scatter and non photopeak events, but you also reduce the number of accepted counts, which increases image noise. This trade off is central in the design of imaging protocols and in performance evaluation.
In GATE, you typically implement energy windows as part of the digitizer configuration for singles. You specify one or more windows by setting appropriate lower and upper values, each in physical energy units. The digitizer then tags or selects singles according to these ranges. For PET, there is often a single main window, for example 350 to 650 keV, that defines which singles are eligible for coincidence sorting. For SPECT, you may configure several windows simultaneously, for example one photopeak window and one or more scatter windows. These windows can then produce different output datasets, such as separate projection images per window.
When you analyze your simulation output, energy windows are an essential reference. For instance, when you plot the energy spectrum of singles, you should clearly see the chosen window superimposed to verify that it covers the intended part of the spectrum. When computing performance metrics, such as sensitivity or scatter fraction, you must always state which energy window definition you used, since the values can change significantly with different window settings.
In summary, lower and upper thresholds are the basic building blocks from which energy windows are constructed. Thresholds control the minimum and maximum accepted energies at the detector signal level, while windows define the specific energy ranges that are relevant for imaging or dosimetry tasks. Correctly chosen thresholds and windows in your GATE digitizer are crucial for realistic modeling of detector behavior and for meaningful comparison with experimental or clinical data.
Views: 10