[MINC-users] Once again: how to use insect properly?

[Alex ZIJDENBOS]

I'd say that the fact that BG and CSF are considered one class is not
(or shouldn't be) related to the tag cleaning stage. I assume that
classify_clean without -clean_tags doesn't use the (separate) BG tag
file, which I would consider a bug.

The reason why clean_tags was removed as a default was because its
performance was never objectively validated. But I suppose we could
change the default back to -clean_tags (sigh), since the overall
consensus appears to be that it improves things...

-- A

On Wed, Apr 19, 2006 at 10:03:01AM -0400, Andrew Janke wrote:
> Myself I always run classify with -clean_tags
> 
> 
> a
> 
> On 4/19/06, Jason Lerch <jason at bic.mni.mcgill.ca> wrote:
> > The easy change is to edit the script, replacing this line
> >
> > Spawn(['classify_clean', $TAL, $CLA]);
> >
> > with
> >
> > Spawn(['classify_clean', '-clean_tags', $TAL, $CLA]);
> >
> > I would recommend using tag cleaning in your classification. The default
> > behaviour has been changed multiple times as we argue amongst ourselves
> > what the default should be. I vote tag cleaning by default. Anyone else?
> > Alex?
> >
> > Jason
> >
> >
> > Ed Gronenschild wrote:
> >
> > >Hi Jason,
> > >
> > >I have done like you proposed and got a correct classification,
> > >or better, CSF was labeled differently from the background.
> > >However, I noticed that the classification in general was different
> > >from that without the option -clean_tags. So two questions remain:
> > >1. which option should I use?
> > >2. how to implement the option -clean_tags in the insect Perl
> > >       script? Should I do this myself or do you have a proposal?
> > >
> > >Cheers,
> > >
> > >Ed
> > >
> > >On 14 Apr 2006, at 18:00, minc-users-request at bic.mni.mcgill.ca wrote:
> > >
> > >
> > >
> > >>Message: 1
> > >>Date: Thu, 13 Apr 2006 12:40:16 -0400
> > >>From: Jason Lerch <jason at bic.mni.mcgill.ca>
> > >>Subject: Re: [MINC-users] Once again: how to use insect properly?
> > >>To: MINC users mailing list <minc-users at bic.mni.mcgill.ca>
> > >>Message-ID: <443E7EF0.8080208 at bic.mni.mcgill.ca>
> > >>Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> > >>
> > >>Ed Gronenschild wrote:
> > >>
> > >>
> > >>
> > >>>Hi,
> > >>>
> > >>>I posted this a week ago and until now I didn't get an answer.
> > >>>So once more:
> > >>>
> > >>>I've used insect to do the classification into WM, GM and CSF. I
> > >>>noticed, however,
> > >>>that the background gets the same class label (=1) as CSF, so I'm not
> > >>>able
> > >>>to distinguish CSF from the background. (WM gets label 3 and GM label
> > >>>2).
> > >>>
> > >>>The command was insect input_volume.mnc, where the volume was a T1
> > >>>coronal stack.
> > >>>
> > >>>
> > >>>
> > >>>
> > >>>
> > >>Can you try using classify_clean -clean_tags on its own on a file in
> > >>stereotaxic space? And, should that not work, can you provide a
> > >>listing
> > >>of your (prefix)/share/classify directory, where prefix is the package
> > >>installation location?
> > >>
> > >>Cheers,
> > >>
> > >>Jason
> > >>
> > >>
> > >>
> > >>>Have I done something wrong?
> > >>>
> > >>>The version of insect is from the classify-1.0 package and I' m on a
> > >>>Mac OSX 10.3.9.
> > >>>
> > >>>Ed
> > >>>
> > >>>
> > >
> > >_______________________________________________
> > >MINC-users at bic.mni.mcgill.ca
> > >http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
> > >
> > >
> >
> > _______________________________________________
> > MINC-users at bic.mni.mcgill.ca
> > http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
> >
> 
> 
> --
> Andrew Janke   (a.janke at gmail.com || http://a.janke.googlepages.com/)
> Canada->Montreal                                   Cell: +1 (514) 924 2012
> 
> _______________________________________________
> MINC-users at bic.mni.mcgill.ca
> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users

-- 
=========================================================================
| Alex P. Zijdenbos, Ph.D.        |                                     |
| McConnell Brain Imaging Centre  |  Phone: [+1] 514 398-5220  (office) |
| Montreal Neurological Institute |         [+1] 514 398-1996   (dept.) |
| 3801 University St., WB-208     |    Fax: [+1] 514 398-8952  (office) |
| Montreal, Quebec, H3A 2B4       |         [+1] 708 810-3309 (private) | 
| CANADA                          | E-mail:      alex at bic.mni.mcgill.ca |
=========================================================================