KAHIBARO
31.4. Energy Resolution
Table of Contents
Energy blurring
In a real gamma camera, the measured energy of each detected photon is never exactly equal to the true photon energy. The scintillation crystal, light collection, photodetectors, and electronics all introduce fluctuations. This limited precision is described by the detector energy resolution, usually quoted at a specific energy, for example “10% at 140 keV” for a Tc‑99m gamma camera.
In GATE, you do not usually simulate all microscopic optical processes when you build a clinical gamma camera model in this part of the course. Instead, you model the limited energy resolution by applying an energy blurring step to the detected events. This is done in the digitizer chain that processes hits into singles for the gamma camera.
Energy blurring replaces the “ideal” deposited energy by a “measured” energy, sampled from a statistical distribution around the true value. The most common choice in GATE for gamma cameras is a Gaussian distribution, because many independent fluctuations (light production, photoelectron statistics, electronic noise) combine to produce a nearly normal distribution of measured energy.
Formally, for a true deposited energy $E_{\text{true}}$, the measured energy $E_{\text{meas}}$ is drawn from
$$
E_{\text{meas}} \sim \mathcal{N}\bigl(E_{\text{true}}, \sigma^2(E_{\text{true}})\bigr),
$$
where $\sigma(E)$ is the standard deviation of the Gaussian at energy $E$. The detector energy resolution is often specified in terms of the full width at half maximum (FWHM) of the photopeak. The relation between FWHM and $\sigma$ for a Gaussian is fixed and very important when you set up the blurring model in GATE.
For a Gaussian distribution
$$
\text{FWHM} = 2.355 \,\sigma.
$$
If the energy resolution is given as a relative FWHM,
$$
R(E) = \frac{\text{FWHM}(E)}{E},
$$
then
$$
\sigma(E) = \frac{R(E)\,E}{2.355}.
$$
In practice for a gamma camera, you will choose a model for $R(E)$, then let the digitizer apply one random Gaussian blur per event.
There are two typical uses in a simple gamma camera simulation in this course. First, you approximate the detector energy resolution as constant across the energies of interest, for example $R(E) \approx R(E_0)$ for all $E$ near the Tc‑99m photopeak at $E_0 = 140 \,\text{keV}$. Second, you use that blurred energy when you later apply an energy window to separate primary photons from scattered ones in the projection data chapter.
Table 1 summarizes how the different quantities are related.
| Quantity | Symbol | Example value |
|---|---|---|
| True photon energy | $E_{\text{true}}$ | $140 \,\text{keV}$ |
| Measured energy | $E_{\text{meas}}$ | random about 140 keV |
| Relative FWHM resolution | $R(E)$ | $0.10$ for 10 % |
| FWHM at energy $E$ | $\text{FWHM}(E)$ | $14 \,\text{keV}$ |
| Gaussian standard deviation | $\sigma(E)$ | $\approx 6 \,\text{keV}$ |
In a GATE gamma camera digitizer, you configure an energy blurring module and tell it what functional form to use for $\sigma(E)$ or for $R(E)$. For a first, beginner friendly model, you might specify a constant resolution at one reference energy and assume the same relative resolution for all events. In that case, for each detected photon, GATE takes the ideal deposited energy from the hits, calculates $\sigma$ from your chosen resolution, then draws a blurred energy value from the Gaussian distribution. The rest of the digitizer chain, including the energy window selection, uses this blurred energy, which makes the simulated energy spectrum look realistic, with a broadened photopeak and a finite ability to reject scattered photons.
Views: 12