[Stir-users] 3D sinograms and voxel units

[Kris T. <kri...@cs...>]

Hi George,

> From: George Kontaxakis [mailto:gk...@di...]
> Sent: 21 November 2002 17:37
> To: Kris Thielemans

> I have a couple of more questions for you:
>
> We have some simulated sinograms from our small  animal  PET scanner (3D).
> In order to feed them  to  STIR,  we obviously need to create an Interfile
> with these data.  I am a bit  puzzled though how to order the 2D sinograms
> (direct, oblique, etc.)  that make up our 3D data set into that
> Interfile.
> Is  there  any  convention  that  uses  STIR regarding the order
> of the 2D
> sinograms into the input Interfile?
>

This question has more or less been answered (but not completely) in a
previous email. I've now made it into a PDF file (attached), using a nice
text that Claire Labb'e wrote for the PARAPET project as starting point. Let
me know what you think about it.

It does not really tackle Interfile, but I sort of hope that this is obvious
once you understood axial compression (and read my Interfile page).

> We are now successfully reconstructing ECAT7 2D sinograms (ECAT EXACT HR+)
> I notice however an artifact at the circle that defines the field of view,
> which gets relatively high values  (as if there were activity  outside the
> field of view) instead of the expected near-zero values.  Any  suggestion?

It's probably there because you are not using a proper model: you subtracted
delayed (so negatives in the data, which after truncation to positives gives
a bowl shaped profile) and probably didn't have scatter correction. Even if
you sorted all that out, you might have a ring when not using matching
forward and back projector. We used to truncate it away, but I removed that
code, as the ring really shows you do not handle the data correctly. (By the
way CTI 7.2 transmission reconstruction has that ring as well, but at very
high magnitude).

> Could you please also confirm  if the voxel values in the output image are
> SUVs?
>
No SUVs no way. uncalibrated ECAT counts if you didn't zoom (i.e. the same
values as you get when usin gthe CTI reconstruction). If you did use zoom,
there might be a scale factor zoom^2 depends on which projectors you used,
not sure about that. sorry.

Hope this helps,

Kris