[MINC-users] mincstats -mask

[Najmeh Khalili M.]

Just two cents worth of request:

If you are going to change anything, please don't forget to write the
manual a bit more trivial.

For example, I don't understand what -mask_floor <min1>,<min2>...<minN> is
trying to acheive. In fact, I always thought in any data set, there exist
only ONE min, but it must be the old school of statistics ;) So I would
appreciate some enlightenment on what subsets these several mins belong
to! (I can assume they corespond to binvalue <val1> <val2> ...<valN> ...
but why should I ASSUME?)

Thanks,
Najma


On Wed, 20 Jul 2005, Alex ZIJDENBOS wrote:

> Actually, perhaps '-binvalue 1' is a bit too restrictive. I've seen
> people create masks with values higher than one (e.g., using
> Display). I think that assuming a "mask" is made up of zero- and
> non-zero values (with a bit of tolerance perhaps for roundoff) would
> be reasonable. But if that's not general enough (I suppose that was
> Peter's argument?), generating an error message when the mask
> treatment is undefined makes sense too.
>
> volume_stats, mincstats predecessor, does the former (without
> tolerance though). If a mask is present but no extrema are given,
> it'll test for mask value != 0.
>
> -- A
>
> On Wed, Jul 20, 2005 at 05:33:38PM -0400, Andrew Janke wrote:
> > Sounds fair to me.
> >
> > Peter and I discussed this at length when we first wrote the thing but
> > never managed to come to agreement.
> >
> > What has just been suggested is what I was hammering for... :)
> >
> >
> > a
> >
> > On 20/07/05, Robert VINCENT <bert@bic.mni.mcgill.ca> wrote:
> > > Hi Alex,
> > >
> > > This appears to be unchanged.  I hadn't been aware of the problem.
> > >
> > > I think adding some kind of default behavior such as -mask_binvalue 1
> > > would not break anything.  But I'm curious to hear other's opinions.
> > >
> > >         -bert
> > >
> > >
> > > On Wed, 20 Jul 2005, Alex ZIJDENBOS wrote:
> > >
> > > > A kind soul just alerted me to something possibly confusing with
> > > > mincstats, being that it will (silently) ignore a -mask argument if
> > > > not given at least one of -mask_{floor,ceil,range,binvalue}.
> > > >
> > > > In other words, the call
> > > >
> > > >    mincstats -mask <mask.mnc> -floor <your_favorite_treshold> <in.mnc>
> > > >
> > > > does nothing with <mask.mnc>. Although this is stated in the man page,
> > > > I suspect it is counterintuitive to many people; it seems to me
> > > > mincstats should either do something sensible with the mask (i.e.,
> > > > default to '-mask_binvalue 1'), or barf when given a mask argument
> > > > which is not used.
> > > >
> > > > I have a vague recollection of this being discussed before, but has it
> > > > been fixed in recent releases of minc?
> > > >
> > > > -- A
> > > > _______________________________________________
> > > > MINC-users@bic.mni.mcgill.ca
> > > > http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
> > > >
> > >
> > > _______________________________________________
> > > MINC-users@bic.mni.mcgill.ca
> > > http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
> > >
> >
> >
> > --
> > Andrew Janke      (a.janke@gmail.com || www.cmr.uq.edu.au/~rotor)
> > Canada->Montreal                                   Cell: +1 (514) 924 2012
> >
> > _______________________________________________
> > MINC-users@bic.mni.mcgill.ca
> > http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
> _______________________________________________
> MINC-users@bic.mni.mcgill.ca
> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
>