KAHIBARO
17.2. Digitizer Chain
Table of Contents
Hits
In GATE, digitization begins with hits. A hit is the raw record of a particle interaction in a sensitive volume, produced by the tracking step of the simulation. Hits contain information such as deposited energy, position, time, particle type, event ID, and volume identifiers, but they are still very close to the underlying physics.
The digitizer chain always starts from one or more hit collections. Each sensitive detector, for example a crystal, photodetector, or scoring layer, typically produces its own hit collection. The purpose of the first digitizer modules is to read these collections and transform the raw interactions into more detector oriented quantities.
A typical first step is energy summation within each sensitive volume. Many particles can interact in the same detector element during a single event. Instead of keeping each interaction separately, the digitizer can group all hits that belong to the same detector element and to the same physical event, then sum their deposited energy and compute a representative time and position. This reduces the data volume and moves one step closer to what the detector electronics would see.
In the digitizer chain configuration, you usually specify which hit collection is used as input, then define a module that performs this grouping and summation. The output of this first digital processing stage is often called preclustered hits or crystal level energy deposits. These are still not what a real acquisition system records, but they form the required input for building singles.
Hits are the only stage in the digitizer chain that directly reflects individual particle tracking steps. Every later stage in the chain combines or transforms this information and no longer represents pure physics steps.
To use hits correctly in a digitizer chain, you must ensure that the sensitive volumes are properly defined, that hit collections are produced for the desired detector components, and that the digitizer modules know which hit collections they should read.
Singles
Singles are detector events that represent the response of a single detector element or channel to radiation. In GATE, the digitizer chain transforms grouped hits into singles by applying models of detector electronics and resolution. Singles are much closer to what a real acquisition system stores as raw data.
The transformation from hits to singles usually follows a series of well defined processing stages inside the digitizer chain. After energy summation, the chain applies energy blurring to simulate finite detector energy resolution, for example Gaussian smearing of the summed energy. It can also apply spatial blurring to represent limited position resolution, and time blurring to represent timing resolution. Next, energy thresholds and windows are applied so that only signals within a chosen energy range are kept, which mimics hardware discriminators.
At this point, each surviving record is a single. It has attributes such as blurred energy, blurred detection time, position, and identifiers for the detector crystal, module, and ring or head. It no longer stores information about each individual particle interaction. Instead, it models the net effect of all those interactions as a discrete detector event.
Conceptually, the digitizer chain from hits to singles can be summarized as:
- Read and group hits by detector element and event.
- Sum deposited energy and choose a representative time and position.
- Apply detector response models energy, spatial, and time blurring.
- Apply energy thresholds and windows to reject unwanted events.
- Output the remaining events as singles.
In a typical configuration, you define these steps through sequential digitizer modules, where the output of one module is the input of the next. The order matters. Blurring almost always happens before applying the final energy window, because the window is defined on the measured, that is blurred, energy.
A single is the basic detector event used for almost all later analysis in emission tomography and detector studies. Its energy and time already include detector effects, and its count rate directly reflects the simulated detector performance.
Because singles play such a central role, the digitizer chain must be tuned carefully. If thresholds are too high, many valid events are lost. If energy or time blurring are unrealistic, any efficiency, resolution, or data driven metric derived from singles will not match measurements.
Coincidences
Coincidences are pairs or groups of singles that occur close enough in time to be considered related. In PET, a coincidence usually refers to two singles, one in each of two detector elements, that are assumed to originate from the same positron annihilation event. The final stage of the digitizer chain, the coincidence sorter, operates on singles to build these pairs.
The coincidence module scans through the stream of singles sorted by time. When it finds two singles whose detection times differ by less than the configured coincidence timing window, for example a few nanoseconds, it pairs them as a coincidence event. Additional criteria such as energy windows on each single, detector separation, or exclusion of particular detector regions can also be applied.
The transformation from singles to coincidences can be summarized as:
- Take the list of singles produced by the earlier digitizer modules.
- Sort singles by detection time if this is not already guaranteed.
- Slide a coincidence window along the time axis.
- For each single, search for partner singles that occur within the time window.
- Apply any additional selection conditions and create a coincidence record for each accepted pair.
The output of this last digitizer stage is a coincidence list. Each coincidence entry typically contains the attributes of both singles, such as detector IDs, positions, energies, and times, as well as derived quantities like the time difference between detections. In PET applications, this list is the main input to further processing, for example line of response construction or image reconstruction.
The coincidence timing window is the critical parameter of the last digitizer stage.
A too wide window increases random coincidences.
A too narrow window rejects valid true events and lowers sensitivity.
Coincidence building completes the digitizer chain. Starting from raw hits, the chain progressively models detector response, applies realistic limitations, and finally produces data structures singles and coincidences that mirror what is recorded by real imaging systems and detector electronics. For imaging simulations, especially PET and sometimes SPECT, configuring this full chain correctly is essential to obtain realistic data for performance studies and reconstruction.
Views: 8