KAHIBARO
32.3. Acquisition Time
Table of Contents
Activity
In SPECT acquisition the activity of your source and the chosen acquisition time together determine how many counts you record and how noisy your projection images will be. Activity is usually expressed in becquerels, so in GATE you will typically use units such as $kBq$, $MBq$, or $GBq$. You will often know the injected or phantom activity from an experimental protocol, and then choose an acquisition time per projection angle to achieve the desired image statistics.
In a SPECT simulation you do not directly model every physical decay that would happen during a real clinical scan. Instead, you choose a number of simulated events and a source activity, and GATE links these to a physical acquisition duration. Conceptually, if the true activity is $A$ and the scan duration is $T$, then the expected number of nuclear decays is $A \times T$. Only a small fraction of these decays will produce photons that pass through the collimator holes, interact in the crystal, fall within the chosen energy window, and are finally recorded as counts in your projection images.
You can control this chain of events in GATE through your source definition, physics list for gamma interactions, collimator and detector geometry, and the digitizer settings such as energy window and thresholds. The acquisition time per projection defines how long the system is assumed to acquire data for each angle. For example, if you simulate a parallel hole gamma camera rotating around a phantom, you might choose 60 projections with 30 seconds per projection. In the simulation you can reflect this by specifying the total acquisition duration and, if you model motion, by making sure that the time each projection is acquired matches your rotation schedule.
It is important to distinguish between activity and the number of primary particles simulated. In practice you will usually choose a convenient number of primaries to reach an acceptable statistical uncertainty in the images. You then relate this number to an effective activity and acquisition time using the definition of activity.
The relation between activity, acquisition time, and the number of decays is
$$N_{\text{decays}} = A \times T$$
where $A$ is the activity in $Bq$ and $T$ is the acquisition time in seconds.
Only a fraction of $N_{\text{decays}}$ becomes detected counts in SPECT due to collimation, attenuation, and detector efficiency.
When you simulate long acquisitions or high activities, make sure that the physics and digitizer settings are appropriate for the expected count rates. For basic SPECT teaching simulations you will usually ignore detector dead time and pile-up, but in realistic system modeling these effects can become important at high activity or long acquisition times.
Counts
Counts are the discrete detector events recorded during the acquisition and are the main observable in SPECT. In the context of simulation, counts correspond to processed detector events after digitization, not raw hits in the scintillation crystal. Each count has attributes such as energy, position on the detector, and acquisition time. When you bin counts into projection images, each pixel or bin contains the total number of counts recorded for that detector element and projection angle.
For a given activity and acquisition time, the total detected counts depend on many factors: the radionuclide energy, collimator geometry, patient or phantom attenuation, scatter, and detector efficiency. In a SPECT simulation, you control these factors explicitly. By increasing acquisition time, you allow more decays to occur and therefore potentially more photons to be detected. As a result, the total number of counts in your projection images grows and the relative statistical noise decreases, since Poisson noise scales like the square root of the number of counts.
In practical simulation work you often explore how counts change when you modify acquisition time. For example, you might run several simulations with identical activity but different acquisition times per projection. For each run you can measure the total counts per projection, the count distribution across the detector, and the noise in regions of interest. This helps you understand the trade-off between scan duration and image quality.
GATE provides time information for each detected event, so you can emulate time binning in the same way as in a real SPECT system. If your gamma camera rotates continuously or in a step-and-shoot pattern, you can assign counts to specific projection angles based on their detection time. The length of each time interval per angle corresponds to the acquisition time for that projection.
For SPECT projections with constant activity, the expected number of detected counts $C$ in a projection is approximately proportional to
$$C \propto A \times T_{\text{proj}} \times \varepsilon$$
where $A$ is the activity, $T_{\text{proj}}$ is the acquisition time per projection, and $\varepsilon$ is the overall detection efficiency of the system.
More counts reduce relative statistical noise, which scales as $1 / \sqrt{C}$.
When you design a SPECT acquisition in GATE, you typically aim for a count level that yields acceptable noise while keeping acquisition times realistic. For absolute beginners an effective workflow is to fix the activity, run a short test simulation, look at the resulting counts and noise in the projection images, then adjust acquisition time and, if necessary, the number of simulated primaries until you obtain stable and interpretable images.
Views: 11