KAHIBARO
31.5. Energy Window
Table of Contents
Photopeak window
In a gamma camera simulation, the energy window defines which detected events are accepted based on their measured energy. The photopeak window is centered around the main gamma energy of the radionuclide and is used to select photons that most likely interacted through the photoelectric effect in the scintillation crystal.
In GATE, you typically configure the energy window as part of the digitizer chain that produces singles from hits in the NaI or other scintillation crystal. Before applying the window, hits have already been summed per crystal and blurred to account for the detector energy resolution. The photopeak window then operates on these blurred energies.
For a radionuclide like Tc-99m, the primary photon energy is about 140 keV. A typical photopeak window is defined as a percentage around this energy, for example 15 percent or 20 percent.
You can express a symmetric photopeak window around a central energy $E_0$ and relative width $w$ (for example 0.15 for 15 percent) as:
$$
E_\text{low} = E_0 \, (1 - w), \quad E_\text{high} = E_0 \, (1 + w)
$$
For Tc-99m, $E_0 \approx 140 \ \text{keV}$ and $w = 0.15$ gives:
$$
E_\text{low} \approx 119 \ \text{keV}, \quad E_\text{high} \approx 161 \ \text{keV}
$$
In a GATE gamma camera simulation, you implement this by configuring the lower and upper energy thresholds in the digitizer. After energy blurring has been applied, any event with measured energy between $E_\text{low}$ and $E_\text{high}$ is kept as a valid photopeak event, and others are rejected or sent to a different output, depending on your setup.
The photopeak window trades sensitivity for image quality. A narrow window (small $w$) reduces scattered photons and improves contrast but also discards more counts, which increases image noise. A wider window collects more events, improving statistics but also admitting more scatter. The optimal width depends on the detector energy resolution and the clinical or research task.
You can summarize the logic of photopeak selection as:
| Quantity | Meaning |
|---|---|
| $E_0$ | Nominal photon energy (for example 140 keV) |
| $w$ | Relative half width (for example 0.15) |
| $E_\text{low}$ | Lower window bound |
| $E_\text{high}$ | Upper window bound |
| Condition to accept | $E_\text{low} \le E_\text{meas} \le E_\text{high}$ |
Photopeak selection rule:
For a given gamma energy $E_0$ and relative window width $w$ (for example 0.15 for 15 percent),
$$
E_\text{low} = E_0 (1 - w), \quad E_\text{high} = E_0 (1 + w),
$$
and an event is accepted in the photopeak window if
$$
E_\text{low} \le E_\text{meas} \le E_\text{high}.
$$
In practice, you should first generate an energy spectrum from your simulated detector events, then visually choose a reasonable window range around the photopeak and implement that range as your photopeak window in the GATE digitizer configuration.
Scatter window
The scatter window is an additional energy window placed below the photopeak to estimate the contribution of scattered photons. Scattered photons deposit less energy in the crystal than unscattered photons, so their energies form a broader distribution below the main photopeak.
In SPECT, scatter correction methods often use counts in one or more scatter windows to approximate the scatter component inside the photopeak window. To support such methods, your gamma camera simulation should record not only photopeak counts but also counts in a dedicated scatter window.
A single scatter window is usually defined just below the photopeak window. For Tc-99m, a common choice is a window that starts some keV below the photopeak lower limit and extends further downward. For example, if the photopeak window is 119 to 161 keV, a simple scatter window might be 90 to 119 keV. The exact choice depends on your scatter correction method and the detector energy resolution.
You can define a scatter window with lower and upper bounds $E_\text{sc,low}$ and $E_\text{sc,high}$, with $E_\text{sc,high}$ below or equal to $E_\text{low}$ of the photopeak window to avoid overlap. In some multi-window methods, scatter and photopeak windows may partly overlap by design, but for a basic gamma camera simulation it is clearer to keep them separate.
In a GATE simulation, you can implement scatter windows in two main ways. You can create separate output branches in the digitizer, each with its own pair of energy thresholds, one for the photopeak window and one for the scatter window, and record counts separately. Or you can record all singles with a wide energy range and later apply software energy windows during analysis, for example in ROOT or Python, by selecting events whose measured energies lie in the desired scatter range.
The role of the scatter window is not to produce the final image but to provide information used in scatter correction. A common conceptual approach is:
- Acquire counts in the photopeak window, $C_\text{pp}$.
- Acquire counts in one or more scatter windows, $C_\text{sc}$.
- Use a model to estimate scatter in the photopeak window as a fraction of $C_\text{sc}$, for example $k \, C_\text{sc}$, where $k$ depends on your window definitions and isotope.
- Subtract the estimated scatter contribution from the photopeak counts.
The details of scatter correction are outside this chapter, but your simulation must provide appropriately configured scatter windows so that these methods can be applied to GATE data.
For beginners, a practical workflow in a gamma camera project is to begin with a single photopeak window to get basic images, then add one scatter window below the photopeak and compare images with and without simple scatter correction, to see the impact of your window choices on image contrast and noise.
Views: 11