STIR 2026 hackathon projects

[KrisThielemans]

Suggestions for the STIR 2026 hackathon that will take place on 5 & 6 March 2026.

For inspiration, see also issues/PRs/discussions in https://github.com/UCL/STIR

Tip

See deepinv/deepinv#700 for examples of how to write a suggestion in this discussion.

[KrisThielemans]

CUDA-managed memory and GPU support

Leads: @casperdcl, @KrisThielemans
Idea: optionally use CUDA managed-pointers for images/arrays to avoid data copies

See #1239 and #1679

Note that the alternative approach in #1583 is very valuable as well, but will not be pursued in this hackathon.

[danieldeidda]

Create a SPECT GPU projector

Lead: @danieldeidda
Idea: create a GPU projector inspired by the rotating projector with gaussian interpolation by Wallis et al, TMI, 1997

PR #1687

[NikEfth]

SimSET–STIR Integration & TOF Scatter Workflow

Leads: @NikEfth, @roHarrison & @rmapree
Idea: A focused hackathon effort to (1) finalize detector‑binning→sinogram conversion for STIR, (2) validate workflows across different scanner geometries, (3) prepare a minimal end‑to‑end SimSET→STIR reconstruction pipeline, and (4) explore TOF‑based scatter estimation using SimSET energy binning

1. list-mode SimSET history file reader testing
   • small scanner
   • Arnau's Quadra simulation (file exists)
2. block detector binning (without and with gap)
   Note: try list-mode reader with lm_to_projdata to compare with sinogram binning
3. generation of SimSET input files from STIR scanner information (Christian and Robert)
4. TOF scatter simulation (in STIR) (Viet)
5. TOF scatter simulation in SimSET (done in Vereos example)
   Note: compare 4 and 5

[ludovicabrusaferri]

SimSET&Normalisation

Lead: @ludovicabrusaferri

normalisation on a Simset simulation, and using it for later quantitative recon

[KrisThielemans]

Quadra conversion script

Lead: Zekai
Fix rotational offset in e7tools output

[markus-jehl]

Reconstruction from PETSIRD data

Lead: @VietAnhDao , @markus-jehl , @eliseemond
Idea: Translate existing STIR raw data to PETSIRD and attempt to reconstruct it in STIR

PR https://github.com/UCL/STIR/pull/1604/files