[Homer-users] HomER- New Release!

Thu Sep 22 08:26:28 EDT 2005

	Sorry it took so long to respond to this (too many deadlines).  960s is a
longer data file then I usually process, but it should be okay.  I've done
300s @100hz (times multiple files simultaniously open) before.
Unfortunatly, it probably is the RAM on that computer that is causing the
issues.

Couple of sugestions:

1)  Resample the data to a lower sample freq.  10Hz is plenty fast to see
the hemodynamic response.  If your first step in analysis is to low-pass
filter the data to 0.5-1hz, then you can resample the data to 2hz without
changing the data.  I would suggest using a good rejection Nyquist filter
when doing this.  The matlab "resample" command will work fine for most
applications.

2)  Averaging has MUCH less memory requirements than deconvolution.  For a
properly designed stimulus presentation (timing), the averaging (i.e.
backprojection) and least-squares deconvolution solution are the same (Dale
1999- HBM 8:109-114).  (You said you were averaging, but something to keep
in mind).

3)  (!!!)  There is a command on the FILE >> pull-down menu called "Clean
Memory".  This will remove the intermediate variables from the dynamic
memory.  For example, when updating it goes Raw data >> HPF >> dOD >> PCA
filters >> FInal result.  CLicking this button removes the variables like
HPF.  You can try this after updating, but before averaging.

I'll see what I can do with optimizing the code for memory usuage.

 Ted Huppert, M.Sc.

PhD student-Harvard Univ.
Dept of Biophysics
Photon Migration Imaging lab
Mass General Hospital/CNY

Tele: (617)726-1223
Cell: (617) 869-1205

thuppert at nmr.mgh.harvard.edu

P.s.  I miss-typed last time.  For the deconvolution of 300s of data at 100
Hz, that's (100 x 300)^2 x 16bit =

	1.8 GB of RAM!!!  (not 14GB: opps- there's 8 bits to a byte and computers
are measured in bytes).  That would still be very taxing on any computer
though.

-----Original Message-----
From: yzhu [mailto:yzhu at nlpr.ia.ac.cn]
Sent: Thursday, September 15, 2005 9:27 PM
To: thuppert at nmr.mgh.harvard.edu; homer-users at nmr.mgh.harvard.edu
Subject: Re: [Homer-users] HomER- New Release!

we use 28 Src-Det channels. wavelengths are 830nm and 690nm.
sample frequency is 10Hz for 960s.
I can process the sample data.
The problem  only appeared when averaging and I didn't preforming
deconvolution.
Thank you very much for your answer.
----- Original Message -----
From: "thuppert" <thuppert at nmr.mgh.harvard.edu>
To: "yzhu" <yzhu at nlpr.ia.ac.cn>; <homer-users at nmr.mgh.harvard.edu>
Sent: Thursday, September 15, 2005 8:50 PM
Subject: RE: [Homer-users] HomER- New Release!

> What is the size of the data?  (i.e. how many Src-Det
> channels/wavelengths/sample frequency/time points )
>
> Were you able to process the sample data??
>
> Was this a problem loading the data, filtering it (updating), or only with
> averaging?  If this was a problem with averaging, were you preforming
> deconvolution (i.e the deconvution check box was checked)?
>
> Deconvolution is the most memory exhaustive process, since it requires the
> inversion of a matrix which is <number of time points > ^2.  For 300s of
> data at 100 Hz, that's (100 x 300)^2 x 16bit = 14GB of RAM!!!   (at 10Hz,
> that's only 150Mb).  I typically use data at 10Hz sample frequencies,
> especially if I am doing event-related stimulus designs, which require
> deconvolution.  Some of the statistics also requires similar inversions.
>
> The memmory usuage I quoted was just to run the program (its what the
Matlab
> Runtime uses)... the actual usuage depends on the size/ammount of data.
>
>
>
> Ted Huppert, M.Sc.
>
> PhD student-Harvard Univ.
> Dept of Biophysics
> Photon Migration Imaging lab
> Mass General Hospital/CNY
>
> Tele: (617)726-1223
> Cell: (617) 869-1205
>
> thuppert at nmr.mgh.harvard.edu
>
>
>
>
>
>
>
> -----Original Message-----
> From: yzhu [mailto:yzhu at nlpr.ia.ac.cn]
> Sent: Wednesday, September 14, 2005 11:21 PM
> To: thuppert at nmr.mgh.harvard.edu
> Subject: Re: [Homer-users] HomER- New Release!
>
>
> Hello,when I run Homer to process my NIRS data,
> a waring dialog which said out of memory appeared .
> When I use averaging--options--(re)Calculate Statistics, it appeared.
> I want to know the reason why it can't operate.
> My OS is win2000, and memory is 1GB.
> ----- Original Message -----
> From: "thuppert" <thuppert at nmr.mgh.harvard.edu>
> To: <homer-users at nmr.mgh.harvard.edu>
> Sent: Thursday, September 15, 2005 2:53 AM
> Subject: RE: [Homer-users] HomER- New Release!
>
>
> > You have the right version.  The new version of HomER runs under the
> Matlab
> > Runtime Component.  This is seperate from the normal Matlab program that
> > most people are familar with.  The Runtime Component is a seperate
> program,
> > which I have supplied with the download of HomER. This program allows
the
> > execution of matlab scripts (*.m files) WITHOUT runing under normal
matlab
> > (i.e. so the computer doesn't need to have the full licensed version of
> > Matlab installed or any toolboxes).
> >
> > After you download HomER and install, check the box to install the
Runtime
> > Component at the end of the installation process.  A seperate
installation
> > package will deploy and walk you through that set-up.  When this is
> > finished, you will find that there is an icon for HomER.exe on the
> desktop.
> > Click on this to start HomER.  HomER will start in its own window (the
> > normal Matlab program is not being run!!).  It takes a few moments the
> first
> > time you run HomER since there is a few house-keeping things to do the
> first
> > time it deploys (all subsequent times will be faster!)
> >
> >
> > -Ted
> >
> >
> > Ted Huppert, M.Sc.
> >
> > PhD student-Harvard Univ.
> > Dept of Biophysics
> > Photon Migration Imaging lab
> > Mass General Hospital/CNY
> >
> > Tele: (617)726-1223
> > Cell: (617) 869-1205
> >
> > thuppert at nmr.mgh.harvard.edu
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > -----Original Message-----
> > From: homer-users-bounces at nmr.mgh.harvard.edu
> > [mailto:homer-users-bounces at nmr.mgh.harvard.edu]On Behalf Of Margaret
> > Duff
> > Sent: Wednesday, September 14, 2005 2:33 PM
> > To: homer-users at nmr.mgh.harvard.edu
> > Subject: Re: [Homer-users] HomER- New Release!
> >
> >
> > Hello,  I downloaded Homer today and at the end it said I needed Matlab
> > Component Runtime to be installed.  But here, it said no Matlab was
needed
> > and nothing would be installed for Matlab.  Did I get the wrong version?
> > Or is the Matlab Component Runtime separate from regular matlab?
Thanks,
> > Margaret
> >
> > On Tue, 30 Aug 2005, thuppert wrote:
> >
> > > We are pleased to announce the newest release of the HomER program for
> > NIRS
> > > data analysis!  This program is now available for download from the
> Photon
> > > Migration Imaging Laboratory at the Massachusetts General Hospital:
> > >
> > > http://www.nmr.mgh.harvard.edu/DOT/resources/homer/home.htm
> > >
> > >
> > > This program is being released in a binary executable format.  This
> newest
> > > version does NOT require any version of Matlab nor any toolboxes to be
> > > installed on your computer.  Also included is a more complete user's
> guide
> > > detailing the new features of the program, as well as full details on
> > > initial installation.
> > >
> > > Note:  This new version is fully compatible with all previous versions
> of
> > > the saved data files.  Refer to the user's guide for more details.
> > >
> > >
> > >
> > >
> >
>
****************************************************************************
> > > **************
> > > We would also like to remind people that we ask for registration
> > information
> > > when you first download HomER.  This information is very important to
us
> > > because we wish track HomER usage for reporting this information to
the
> > NIH,
> > > which funds this work.  We therefore ask that you be as accurate as
> > possible
> > > when filling out these forms.
> > >
> > >
> > >
> > >
> > > Questions and comments can be submitted to the
> > > homer-users at nmr.mgh.harvard.edu email list server.
> > >
> > >
> > >
> > > Sincerely-
> > >
> > >
> > > Ted Huppert, M.Sc.
> > >
> > > PhD student-Harvard Univ.
> > > Dept of Biophysics
> > > Photon Migration Imaging lab
> > > Mass General Hospital/CNY
> > >
> > > Tele: (617)726-1223
> > >
> > > thuppert at nmr.mgh.harvard.edu
> > >
> > > _______________________________________________
> > > Homer-users mailing list
> > > Homer-users at nmr.mgh.harvard.edu
> > > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/homer-users
> > >
> > >
> > >
> > _______________________________________________
> > Homer-users mailing list
> > Homer-users at nmr.mgh.harvard.edu
> > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/homer-users
> >
> >
>
>
> --------------------------------------------------------------------------
--
> ----
>
>
> > _______________________________________________
> > Homer-users mailing list
> > Homer-users at nmr.mgh.harvard.edu
> > https://mail.nmr.mgh.harvard.edu/mailman/listinfo/homer-users
>
>
> _______________________________________________
> Homer-users mailing list
> Homer-users at nmr.mgh.harvard.edu
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/homer-users
>