Forward projection for CTAC issue - help welcomed!

[kmo49]

Hello!

I am trying to forward project a three dimensional CT object to use for attenuation correction of my PET image. I am modeling 3D PET with 24 rings, maximum segment of 23. My resultant PET sinogram is 249x210x576. 210 aa bins and 249 td bins.

When I am forward projecting the CT image, the sinogram looks correct on the first slice and the last slice, but proceeds to look like stacked CT images towards the middle.

I am getting the same issue with the calculate_attenuation function. I included images below, and would appreicate any advice on how to fix this issue, thank you!

[KrisThielemans]

I suspect you are reading the data wrongly. Are you doing this without STIR, e.g. read it as a "stack of sinograms"? In that case, it is incorrect. STIR currently writes the data in a different order, check the .hs file. This is fairly difficult to read unfortunately without STIR, so I recommend using STIR python

proj_data = stir.ProjData.read_from_file('bla.hs')
proj_data_arr = stirextra.to_numpy(proj_data)

this array is a stack of sinograms. First seg 0, then +1, then -1, etc, see

STIR/src/include/stir/ProjData.h

Lines 257 to 263 in 42aebde

	//! Return a vector with segment numbers in a standard order
	/*! This returns a vector filled as \f$ [0, 1, -1, 2, -2, ...] \f$.
	In the (unlikely!) case that the segment range is not symmetric,
	the sequence just continues with
	<i>valid</i> segment numbers, e.g. \f$ [0, 1, -1, 2, 3 ] \f$.
	*/
	static std::vector<int> standard_segment_sequence(const ProjDataInfo& pdi);

or similar lines in SIRF python.