Patient-like phantoms

To study how a camera performs on a clinically realistic task, you need a source that looks like a patient: tumor uptake in the liver, physiological uptake in the kidneys and salivary glands, marrow activity, and metastases, each attenuating and emitting as the real tissue would. This package gives you two ways to build such a source and image it through any collimator and detector from the design chapter.

For quality-assurance phantoms (Jaszczak, NEMA IEC body, NEMA NU-1 resolution sources) used to assess system accuracy, see QA phantoms.

The phantoms use only native OpenTOPAS components (G4Ellipsoid, TsSphere, volumetric sources), so they are self-contained parameter content that runs in any OpenTOPAS deck, whether you are doing SPECT imaging or dosimetry-methods work. The uptake-ratio evidence behind the presets lives separately in the lab knowledge base, so the presets stay portable.

Composable analytic phantoms

A Lu-177 PSMA scenario: kidneys, liver, spleen, salivary glands, spine, and a nodal metastasis, with the soft-tissue torso hidden.

Build a phantom from native shapes, each carrying its own activity. See phantoms/README.md for the full recipe: the geometry building blocks, the volumetric-source pattern, and the weighting rule that sets each region's history count in proportion to concentration × volume.

Four ready scenario presets, with literature-sourced uptake ratios, are in phantoms/scenarios/:

Preset	Agent	Regions and notes
lu177_psma.txt	177Lu-PSMA (mCRPC)	kidneys, salivary, liver, spleen, spine marrow, plus nodal and bone metastases; 208 keV
ac225_psma.txt	225Ac-PSMA (with daughters)	the same parent layout as 177Lu at 218 keV (221Fr), plus a 213Bi kidney-redistribution source at 440 keV (the 440/218 kidney signature)
y90_sirt.txt	90Y radioembolization	liver: perfused normal, tumor (T:N ≈ 4), and lung shunt (~5%); β particles produce bremsstrahlung, imaged at ~90–110 keV
i131_mibg.txt	131I-mIBG (neuroendocrine)	liver, heart, salivary, adrenal, spleen, bladder, plus a hot tumor; 364 keV

Each preset is an IncludeFile fragment that defines the body, the organs and lesions, and one weighted source per hot region. Combine it with a collimator, a detector, and a scorer, as in examples/phantom/lu177_psma_projection.txt. Define the acquisition side inline in the deck, since OpenTOPAS rejects two parallel top-level IncludeFile chains.

Where the uptake ratios come from

The uptake ratios are sourced in the workspace knowledge base research/knowledge/rpt-biodistribution/ (not shipped in this repo): 177Lu-PSMA from Violet 2019, Okamoto 2017, Delker 2016, Kabasakal 2017, and Hohberg 2023. The literature reports cumulated dose (Gy/GBq) far more often than an instantaneous 24 h concentration, so the presets use cumulated-dose ratios as relative time-integrated activity, which means a single SPECT image will show lower tumor- and salivary-to-kidney contrast than the ratios alone suggest. The literature gave no background muscle or blood uptake, so the body is a pure attenuator with no background emission; add a low-level background source if your study needs one. The 225Ac, 90Y, and 131I scenarios still await their own sourced passes.

Voxelized XCAT phantoms

For full anatomical realism, use the native OpenTOPAS voxel pipeline, which needs no code from this package:

• Attenuation. TsImagingToMaterialXcat converts an XCAT attenuation image into a voxelized material geometry. Set s:Ge/Phantom/Type = "TsImagingToMaterialXcat" pointing at the XCAT files; it assigns a material per XCAT tissue id.
• Activity. Emit from a matching XCAT activity image with TsDicomActivityMap (or a distributed source that reads the activity image), so the source strength follows the image voxel values.

You supply the XCAT attenuation and activity images, which are licensed separately and not shipped here. Then add a collimator, detector, and scorer as for the analytic phantoms. This path trades the analytic phantom's easy parameter sweeps for full anatomical realism.

Imaging a phantom

A phantom is only the source. To take a single view, add a collimator and detector and a scorer, as in the worked example above. To acquire a full study, in which the detector moves and the isotope decays over the acquisition time, build one deck per view (examples/phantom/spect_acquisition.txt). Because a collimated camera detects a tiny fraction of the emitted photons, patient-like runs are count-starved; see Variance reduction for the forced-detection fast path that makes them tractable.