[gate-users] Crystal ID and position of Singles in PET

[Chen, Yu]

This is a follow-up to my message. 

First Thank you Ange for your suggestion. 

After my further investigation, I found there is no
problem in GATE in matching its crystalID and global
position for singles.  

The discrepancy occurs when I use a perfect ring,
which has a complete circular symmetry, to describe my PET setting
and to convert the position to rsector, crystal IDs since
the real PET setting (in my case, Philips Allegro) only has 
block (rsector) circular symmetry. 

Unfortunately, in many reconstruction methods, we assume a complete
circular symmetry to simplify the calculations. This would lead to 
an extra uncertainty in image reconstruction (equivilent to an angular
blurring), which is scanner geometry-dependent. In my case, crystal ID 
difference is ~20% and only -1 or +1 which corresponds to a maximum
4mm shift.

Be aware of this effect when you evaluate your method. 



Yu Chen, Ph.D.
University of Massachusetts Medical School
Division of Nuclear Medicine
55 Lake Avenue North
Worcester, MA 01655-0243
Phone:  (508) 856-6123
Fax:    (508) 856-4572



-----Original Message-----
From: Angela M K Foudray [mailto:afoudray at stanford.edu]
Sent: Tue 3/14/2006 6:21 PM
To: GATE feedback and helpline for Users
Subject: Re: [gate-users] Crystal ID and position of Singles in PET
 
Hi Yu,

Have you looked at the hits file for the same simulation?  My guess is
that the Compton interactions in the detector and their corresponding
energy blur are making the global position and the crystal ID possibly
not line up.  I don't know how the crystalID is calculated for a single
(since the single can be made of multiple hits with their own crystal
ID) - for instance, the x,y,z position of a single could potentially not
end up in any crystal at all in a system with gaps between crystals...

So maybe looking in the hits file will help track things down?

Ange

On Tue, 2006-03-14 at 16:36 -0500, Chen, Yu wrote:
> Hi, Gaters,
> 
> When I try to convert the global position (globalPosX, Y, Z) of Singles in PET
> to corresponding rsectorID and crystalID, I found there is discrepancy between
> converted crystalID and recorded crystalID, namely, ~10% with -1 (converted - output)
> distributed within lower half of the same rsector while ~10% with +1 in upper half.
> Here lower, upper are defined along anti-clock-wise (increased azimuthal direction in
> X-Y plane).
> 
> Has anybody investigated this before?
> 
> I am using cylindricalPET geometry and setDepth 1, which means Singles are readout from 
> rsector with all pulses in all crystals in that rsector. A singles will take that pulse's
> crestalID and position with max energy. So expected discrepancy should be very small
> due only to float calculation. Am I right? If yes, what's wrong here?
> 
> Thanks,
>   
> 
> Yu Chen, Ph.D.
> University of Massachusetts Medical School
> Division of Nuclear Medicine
> 55 Lake Avenue North
> Worcester, MA 01655-0243
> Phone:  (508) 856-6123
> Fax:    (508) 856-4572
> 
> _______________________________________________
> gate-users mailing list
> gate-users at lphe1pet1.epfl.ch
> http://lphe1pet1.epfl.ch/mailman/listinfo/gate-users
-- 
Angela M K Foudray <afoudray at stanford.edu>


-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 4455 bytes
Desc: not available
URL: <http://lists.opengatecollaboration.org/mailman/private/gate-users/attachments/20060317/54481dea/attachment.bin>