Frequenty asked questions

Relating to image manipulation:

  • I have a NIfTI image and I want to generate the corresponding image parameters JSON file

    • With Python, this one-liner: yrt.ImageOwned("<my_image.nii>").getParams().serialize("<my_params.json>")

      • Without Python, this can be done manually quite easily. The only caveat is that the image parameters file needs to specify an offset, which is the position of the center of the image. This is different from the origin of the image, which is the physical position of the first voxel.

        • \(p_{0_x} = o_x - \frac{l_x}{2} + \frac{v_x}{2}\) where \(p_{0_x}\) is the origin, \(o_x\) is the image offset, \(l_x\) is the image length and \(v_x\) is the voxel size in the X dimension. The same equation applies in the Y and Z dimensions

Relating to CASToR:

  • I have scanner parameters from CASToR and want to use them in YRT-PET

    • For the Look-Up-Table:

      • CASToR provides a tool (castor-scannerLUTExplorer) that allows one to display each detecting element

      • The following command line will generate a log file containing a text description of each detector: castor-scannerLUTExplorer -sf <my geom file> -g -o <output log file>

      • Then, the script in scripts/data_conversion/convert_CASToR_to_YRT-PET_LUT.py takes that log file and converts it into a YRT-PET-ready Look-Up-Table.

    • For the Scanner parameters JSON file:

      • Most of the scanner parameters can be determined manually. Though some caveats are worth mentioning:

        • Note that the minimum angle difference (named minAngDiff in YRT-PET’s JSON file and min angle difference in CASToR’s hscan file) is described in YRT-PET in terms of number of detector elements while in CASToR it is described in degrees.

      • Similarly, the maximum ring difference maxRingDiff property is defined in terms of the number of rings.

  • I have a List-mode file in the CASToR format, how do I convert it in the YRT-PET default ListMode format (LM) ?

    • When there is no Time-of-Flight (TOF) and no Attenuation correction, the CASToR list-mode datatype is equivalent to YRT-PET’s.

    • When there is TOF in the List-mode, CASToR’s structure encodes the TOF slightly differently. The script in scripts/convert_CASToR_to_YRT-PET_list-mode.py does the conversion.

      • Note that the script does not handle normalisation information (CASToR 3.2).

      • YRT-PET’s default List-Mode format does not encode as many fields as CASToR’s, so a one-to-one equivalence is not to be expected.

  • My images seem to be flipped in the Z direction compared to CASToR…

    • This might be because CASToR uses a left-handed coordinate system while YRT-PET uses a right-handed coordinate system.