[Gate-users] RE: Z-grid spacing problem

[Kris Thielemans]

Hi Hyejoo

no need to send email to my address directly. I prefer not to, also to avoid
address-harvesting by the spammers via the email archives.

> 
> I am having a problem with z grid spacing for using STIR.  I 
> read through STIR-user messages but it is not clear to me 
> what I have to change to make my current interfile work for 
> STIR.  When I tried to run STIR with my current .hv file I 
> have gotten the following message
> 
> 
> ERROR: DataSymmetriesForBins_PET_CartesianGrid can currently 
> only support z-grid spacing equal to the ring spacing of the 
> scanner divided by an integer. Sorry
> 
> 
> 
> I was told that changing the value of scaling factor 
> (mm/pixel) [3] from
> 1.27999 to 4.25 should work.  I tried and actually I did not get the error
message.  

great

> I was going to recalculate the matrix 
> size to be 6 or 7 to make scaling factor to be 4.25 
> (21*1.27999/4.25). However, there seems to be certain ways to 
> assign those parameters according to the messages about 
> z-grid in the user message instead of simply fixing the value 
> of scaling factor (mm/pixel) [3] to be 4.25.  
> 


I strongly recommend for an N-ring scanner to use an image with an
odd-number of planes (usually 2N-1) with z-voxel-size =
scanner-ring-distance/2.

The projector will not complain if you use other integer factors (in
particular z-voxel-size=scanner-ring-distance, not so sure about using e.g.
one third the ring spacing) but I believe you can get funny (i.e. wrong)
rounding effects then (e.g. if the number of images is equal to the number
of rings for span=1, I never tried for span=3). 

STIR 2.0 alpha should be a bit better than 1.4 for this, but as far as
reconstruction goes, there doesn't seem to be a huge advantage in providing
this flexibility, so I won't invest any time in this (other can of course).

Of course, you cannot simply change the header, you'll have to generate your
image with a different voxel size (or re-interpolate, zoom_image would help
there). The tricky thing is to make sure that your image is still in the
correct position w.r.t. to the scanner. Currently, STIR assumes that the
image and scanner are centered in axial direction (remember the problems
with even-sized images in x,y, for instance mentioned on SimSET).

Hope this helps

Kris