[Gate-users] Too many randoms in PET NECR simulation

[Dávid Völgyes]

Hi,

It's hard to answer the question in general, but if there is nothing
extraordinary in the scanner setup:
- the number of randoms should be linear function of the coincidence
window width
  (and both true and delayed window should have the same time window)
- randoms should be quadratic function of the activity ( singles are
linear in activity, and randoms are proportional with the product of
the single rates (const*S1*S2))
(As long as the pileup (deadtime) probability is low, but below the
NECR peak this should be more or less true.)

Be careful with the delayed window: it should be far enough from the
true window, otherwise both the random and the true events can be
registered as random. (E.g. if the scanner has cca. 5nsec time
resolution, then use at least 50 nsec delay for the delayed window.)

Check the properties of your random events.
- is it linear in time window width?
- is it quadratic in activity?
- is the time window correct?
- not to blame the manufacturer, but are these also true for the
manufacturer data?
Well, I think that you cannot get four "yes", otherwise your
simulation would agree with the device data.
(You should check the part where you get "no".)

Also you should check your deadtime / pileup model if it is a good
enough approximation of the real device.

David


On Fri, Oct 24, 2014 at 9:57 PM, Leo Read <leoread at gmail.com> wrote:
> Dear Gate Users,
>
> I'm simulating a noise-equivalent count rate curve in a PET scanner. The
> general trends for Trues and Scatter events are good. However, the random
> coincidences are increasing much more rapidly than what has been published
> for this scanner. As one might expect, this leads to a much poorer peak NECR
> than what is expected.
>
> It seems this question has been asked multiple times on the forum, without
> any responses [for example, Benedikt Rossboth asked a similar question on
> 2012-03-07 and Heesoon asked a question on 2013-11-26].
>
> If somebody knows the reason why this might be happening, please share it
> with the forum. Also, please let me know if you have any suggested
> experiments to try for debugging purposes.
>
> Cheers!
>
> _______________________________________________
> Gate-users mailing list
> Gate-users at lists.opengatecollaboration.org
> http://lists.opengatecollaboration.org/mailman/listinfo/gate-users