Skip to content

Information for contributors of phantom data for PETRIC

Kris Thielemans edited this page Jul 2, 2024 · 2 revisions

Information for contributors of phantom data for PETRIC

V1.0.0

Supported data

We can only handle data that are supported by STIR and therefore SIRF, either directly or through one of our conversion scripts. The latter are currently not yet publicly available due to pending approval from scanner vendors.

Currently confirmed scanners for which we accept data:

  • Siemens mMR (native STIR)

  • Siemens Vision 600 (conversion via e7tools)

  • GE Signa PET/MR (conversion via Duetto/pettoolbox, but we have no access to the MRAC code)

  • GE Discovery 690/710 (conversion via Duetto/pettoolbox)

  • GE Discovery MI (conversion via Duetto/pettoolbox)

  • Mediso AnyScan (conversion will be handled by NPL)

  • Positrigo NeuroLF (from output in SAFIR)

This list can be expanded, so other data for testing are welcome.

Phantom information

  • One single bed position, static (no bed or phantom movement nor dynamics).

  • “Interesting” phantom with inserts/regions of different contrast. Non-piecewise continuous data preferred, but not available by most.

  • Any radionuclide

  • Clinically relevant count-level (which could be low, e.g. for Y90)

Required data

Please organise in different folders as follows:

  • README.md
    Textual description of data with as much detail as possible. Please use the skeleton from the appendix.

  • list-mode/
    List-mode data (preferred for longer acquisition than what will be used for reconstruction in the challenge, but not required) as exported from the scanner console

  • AC/
    CTAC or MRAC (in DICOM, Nifti or STIR Interfile format)

  • Calibration/
    Normalisation/calibration files (in the format supplied by the vendor)

  • vendor_recon/
    Image for appropriate time-frame reconstructed with vendor software (DICOM) (all corrections applied, exact reconstruction settings that you use are not important, but should be reported)

  • VOIs/
    VOIs need to be supplied if at all possible. Ideally they are given as images in exact same dimensions/voxel-size as the scanner reconstruction (Nifti or STIR Interfile format), one file for each VOI:

    • VOI_whole.*: roughly whole phantom (can be a bit larger/smaller)

    • VOI_background.*: background VOI (in region of uniform activity (strictly positive))

    • VOI_1.*: interesting VOI 1

    • VOI_2.*: interesting VOI 2

    This can be done in ITKsnap or similar, or programmatically via SIRF, e.g. as in the contributed example for how the VOIs for the NEMA IQ phantom can be constructed.

License

Data will need to be open access. We highly recommend to use the Creative Commons Attribution 4.0 International license.

Uploading

Please contact Harry Tsoumpas [email protected] and Kris Thielemans [email protected] by email for instructions.

After some checks by our team, please upload this data to Zenodo or similar (not urgent, but it will allow us to reference you). See the SyneRBI community on Zenodo for existing data (older data-sets do not follow the description in this document).

We will also upload the “prepared” data to Zenodo, referring to your DOI.

Appendix 1: README.md skeleton for description of phantom data

Please use markdown format (which is just text). Ideally add a photo and/or drawings of the phantom/acquisition set-up in the “root” folder (do not use spaces in filenames).

# Overall description

Some text here

# Phantom description

Some text here

\![alt text\](PhantomPicture.png "Some Title")

# Acquisition information

## Institution

Some text here on where acquired, optionally including names of
personnel.

## Scanner model

(please be specific as you can, model, software version, …)

## Acquisition Date

Use format 19 MAY 2024

## Radiopharmaceutical and nuclide information

e.g. FDG, F-18

## Preparation protocol

Some text on set-up, ideally including expected activities in various
inserts (cross-calibrated to start of PET scan)

## Acquisition protocol

Some text on how the scan was performed, e.g. CT was acquired with
helical CT, keV, mA, rotation-speed, pitch, ... A 40 min PET scan was
performed in list-mode.

## Reconstruction settings

Some text on what the reconstructions settings were (e.g. OSEM, 3
iterations, 7 subsets, post-filter (2D Gaussian of FWHM, and "standard"
z-filter)

## VOI description

Some text on how VOIs were determined etc

### VOI_whole: whole phantom

Something here

### VOI_background: background

Something here

### VOI_1: white matter region

Something here