[gate-users] sample image from my IQ_phantom simulation

Schmidtlein, Ross/Sloan-Kettering Institute schmidtr at mskcc.org
Fri Sep 16 18:32:49 CEST 2005

Thanks Kris for the update.  I'll pass along the modified routines.

What I meant to say, and that you picked up on, is that while the
voxelized input for GATE is not quite in a format that STIR can use, it
is quite close.  My feeling is (and as you suggest) that creating a
translation to an interfile format that STIR can use for attenuation
correction will be fairly straight forward (GATE uses Hounsfield units
and a material range translator to build a voxelized phantom).  In the
previous email I meant to suggest that STIR's image_generation routine
can provide some insight to creating an independent voxel translator
(i.e. coordinate system convention and the like) and should be examined.
The STIR image generation program is very good for regular geometric
shapes such as phantoms and the like and is probably best for those


-----Original Message-----
From: gate-users-bounces at lphe1pet1.epfl.ch
[mailto:gate-users-bounces at lphe1pet1.epfl.ch] On Behalf Of Kris
Sent: Friday, September 16, 2005 10:02 AM
To: 'GATE feedback and helpline for Users'
Subject: RE: [gate-users] sample image from my IQ_phantom simulation

Hi Ross

> Complex phantoms like the IQ phantom may require you to modify the 
> STIR image_generation software.  In it's standard form STIR can 
> generate ellipsoids and ellipsoidal cylinders.
> I have modified some of this code to produce sectors of cylinders and 
> boxes.  I have given these modifications to Kris but they have not 
> been implemented in a STIR release yet.  I am hoping that they are 
> clean enough that he can implement them in the next release with out 
> too much fuss.

At present, your stuff and some other contributions to STIR are under
review by a lawyer. GE doesn't want to put source code on our web-site
for which it isn't absolutely sure that it is allowed to do so...
Nothing stops you from distributing it yourself obviously, although in
the future I sincerely hope I can put it in.

> These modifications should work just as well for a voxel phantoms but 
> may be quite tedious to build.  For voxelized phantoms look at the 
> STIR software and then see if you can translate the GATE voxel 
> matrices into a STIR image it should be very straight forward.  STIR 
> is written cleanly enough that the output from the image_generation 
> routine is easy to follow.

Obviously, if you have a voxelised phantom output by GATE, it would be
far easier to immediately use that as input for STIR. If you have a file
with only the binary data (potentially with an initial header), you can
easily cook-up an interfile header for it.

Hmm. I lost the plot here I guess. Maybe you want to use STIR's
generate_image to have a nice phantom, and use that as input of GATE.


gate-users mailing list
gate-users at lphe1pet1.epfl.ch

More information about the Gate-users mailing list