output of DigitizerProjectionActor

[CIRCLE2096]

I’m using the DigitizerProjectionActor to simulate a CT scanner, but I’m confused about the mhd/raw output.

I’ve looked at the example in Test028 for the SPECT system, where the DigitizerProjectionActor is used after the DigitizerEnergyWindowsActor (as seen in the add_digitizer_tc99m function). The DigitizerEnergyWindowsActor produces several output keys, such as PostPosition_X, TotalEnergyDeposit, and GlobalTime.

However, the raw file output from the DigitizerProjectionActor is a matrix with a shape like (500, 500, 1), containing just a single float value per voxel. I couldn’t find a clear explanation of this value in the documentation, and after reviewing the source code, I only found that the DigitizerProjectionActor works by mapping the hit positions to image indices.

The image seems to be displayed as a grayscale image, so I’m wondering how that value is computed. Pls help.

[optui]

I'm no expert and don't have experience with the SPECT system setup, but here's my two cents:

The DigitizerProjectionActor internally manages data as follows:

• At the start of the simulation, it allocates one 3D float image, with shape [Nx, Ny, Nslices]

  • Nx, Ny -> set by the size parameter (e.g., 500×500).
  • Nslices -> computed as nr. of input digitizer collections × nr. of timing intervals.
    • e.g. if you have 1 DigitizerHitsCollectionActor and 180 timing intervals, you get 180 slices.
  • Source code

• During the simulation, for each hit (digi):

  • The PostPosition is mapped onto 2D pixel indices, based on your chosen spacing, size, and origin_as_image_center parameters.
  • The float value at that pixel is incremented by the digi’s weight (default weight is 1.0).

• At the end of the simulation, the resulting 3D image volume (stack of 2D projections) is saved as a single .mhd/.raw pair:

  • The .mhd file contains metadata
  • The .raw file contains the raw float values, one per voxel (counts per pixel).

So, each voxel contains the sum of digi weights - which corresponds to the number of gamma events if no weighting is applied - for that pixel and projection interval.

---

That’s just my understanding from reading the docs and source code and tinkering a bit - if anyone with more experience spots anything off, or has a more accurate explanation, I’d really appreciate any corrections or insights!

[dsarrut]

exact! One additional information: if you use energy windowing (like in spect), you will get one additional slice per energy windows. Say you have 3 energy windows and 100 rotation angles, the final image (mhd/raw) will be 300 slices.