Memory allocation/speed of processing

[Shannon, Michael]

Hi there

We use imageJ for certain image analysis techniques, in which each frame of
a TIFF containing 30000 frames is processed.

The specific plugin I am using is called 'ThunderSTORM'

Our problem is that this takes an inordinate amount of time to do.

We recently upgraded our computers, so now this one has 48 gb RAM, and its
original intel xeon CPU (e5-1620 with 3.7GHz).

I have upped the memory allocation in ImageJ - is there any other way to
make this go faster?

Thanks for any help anyone can offer,

Mike?


--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[lechristophe]

Hi,

You can also check that the number of cores used by ImageJ (in the ImageJ
options) is optimal.

However, the speed of ThunderSTORM processing is likely due to the plugin
itself, i.e. the way it is coded, and the parameters you use for detection
and localization of blinking events. For example, the multiple-emitter
fitting option is extremely demanding and will slow processing considerably.

I suggest you compare with another STORM software (such as Octane,
QuickPALM in ImageJ, or rapidSTORM) to see if there is a big difference in
processing time with the same parameters (be careful it is sometimes
difficult to translate parameters from one soft to another).

In any case I suggest you ask more question on the ThunderSTORM google
group:
https://groups.google.com/forum/#!forum/thunderstorm-users


Christophe

--
Christophe Leterrier
Researcher
Axonal Domains Architecture Team
CRN2M CNRS UMR 7286
Aix Marseille University, France



On Thu, Feb 19, 2015 at 12:24 PM, Shannon, Michael <
[hidden email]> wrote:

> Hi there
>
> We use imageJ for certain image analysis techniques, in which each frame of
> a TIFF containing 30000 frames is processed.
>
> The specific plugin I am using is called 'ThunderSTORM'
>
> Our problem is that this takes an inordinate amount of time to do.
>
> We recently upgraded our computers, so now this one has 48 gb RAM, and its
> original intel xeon CPU (e5-1620 with 3.7GHz).
>
> I have upped the memory allocation in ImageJ - is there any other way to
> make this go faster?
>
> Thanks for any help anyone can offer,
>
> Mike?
>
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[vbindokas]

Hi Mike,
What do you consider "inordinate" times?
     We run ThunderSTORM on far lesser machines with very good
performance [12GB RAM, 8 threads, 2.8GHz x3460] on 100k frames of 3-D
data. The plugin does slow appreciably if there are very dense SMLs to
be fit per frame and/or you are asking for multiple peak testing/fitting
as Christophe mentioned in his reply.  You might want to compare results
with the multi-emitter feature turned off, with it set to 2, to 3, etc.
Just because it can fit 5 peaks, doesn't mean you should start with
that, for example.
     Fitting data in RAM is a bit faster than from virtual stack, but
virtual mode seems just fine and allows other operations to be run
during the wait time.
regards,
Vytas

On 2/19/2015 5:24 AM, Shannon, Michael wrote:

> Hi there
>
> We use imageJ for certain image analysis techniques, in which each frame of
> a TIFF containing 30000 frames is processed.
>
> The specific plugin I am using is called 'ThunderSTORM'
>
> Our problem is that this takes an inordinate amount of time to do.
>
> We recently upgraded our computers, so now this one has 48 gb RAM, and its
> original intel xeon CPU (e5-1620 with 3.7GHz).
>
> I have upped the memory allocation in ImageJ - is there any other way to
> make this go faster?
>
> Thanks for any help anyone can offer,
>
> Mike?
>
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

--
__

Vytas Bindokas, Ph.D.
Research Assoc. (Assoc. Prof.),
Director, BSD Light Microscopy Core Facility
phone: [phone]

      [address for letters ONLY (see shipping addr below):]
Dept Pharmacol Physiol Sci MC0926
947 E 58th Street
The University of Chicago
Chicago IL 60637
Room Abbott 129


shipping address (main KCBD site):
V. Bindokas
900 E 57th Street
KCBD room 1250, Microscopy Core
The University of Chicago
Chicago IL 60637


email [hidden email]
web site for LMCF:
http://digital.uchicago.edu/index.html

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html