Hi Michael,
The -M option in avwstats gives you the mean only over the non-zero voxels.
This seems to be what you want.
Cheers,
Mark
Lipton, Michael wrote:
> Thanks Mark.
>
> Now that I have an image restricted to my cluster of interest, I have
> collapsed the max and min over time (-Tmax, etc.) This seems to yield
> an image with a maximum value for each voxel. How can I get the mean
> of these maximum values WITHOUT including all of the 0 value voxels in
> the calculation??
>
> Ultimately, I am interested in determining the mean signal change over
> my entire time series for each cluster.
>
> Thanks again,
>
> Michael
>
> > -----Original Message-----
> > From: FSL - FMRIB's Software Library
> > [mailto:[log in to unmask]] On Behalf Of Michael L Lipton
> > Sent: Sunday, August 31, 2003 10:29 PM
> > To: [log in to unmask]
> > Subject: Re: [FSL] FSL View and Clusters
> >
> >
> > Thanks Mark. Once I load that cluster_mask_zstat image into
> > FSLView, how do I determine the integer identity of each
> > cluster? When I click on a cluster does it show this value
> > somewhere in FSLView or do I have to determine the cluster size?
> >
> > Thanks again,
> >
> > Michael
> >
> > On Sunday, August 31, 2003, at 04:49 PM, Mark Jenkinson wrote:
> >
> > > Hi Michael,
> > >
> > > It isn't the thresh_zstat image that you want - that contains the z
> > > statistic values (post-thresholding). What you want is the
> > > cluster_mask_zstat images.
> > > These contain values from 0 to N where N is the number of clusters
> > > found.
> > > Note that the largest cluster has value N, the next largest
> > has value
> > > N-1, etc.
> > >
> > > Then, to generate an image of the cluster number M just use avwmaths
> > > like:
> > > avwmaths cluster_mask_zstat1 -thr M -uthr M -bin outputclustermask
> > >
> > > All the best,
> > > Mark
> > >
> > >
> > > On Friday, August 29, 2003, at 07:57 pm, Michael Lipton wrote:
> > >
> > >> At the course in LA, one of the FMRIB people indicated that when
> > >> viewed in FSLView, each cluster in thresh_zstat.hdr will have a
> > >> unique integer value. This could be used to create a mask of one
> > >> cluster only by thresholding in avwmaths. Now as I look at
> > my zstat
> > >> image in FSLview I am
> > >> not clear how I identify the clusters.
> > >>
> > >> In any case, what is the best way to generate a binary
> > mask of only
> > >> one cluster?
> > >>
> > >> Thanks,
> > >>
> > >> Michael
> > >
> >
>
|