Morphometric statistics problem

Questions about BrainVisa usage and installation

Moderators: denghien, riviere

Post Reply
lewishou
Posts: 8
Joined: Fri Nov 16, 2012 5:30 pm

Morphometric statistics problem

Post by lewishou »

Hello all, my name is Lewis and I'm working in a lab at Edinburgh, Scotland. I've been having a problem with extraction of morphometric statistics from my MR images. When I extract the stats, all my values are 0. The strange thing is that I have previously successfully extracted the stats from a smaller subsample without any problems, but this time I keep getting zeros for all values (see right hand side of screenshot - sorry for the quality, size limits...) despite running the same protocol. I have updated the mac BV to 4.2.1 since the last time, but the same thing happens when I run the linux version of BV on the data and this one has not been changed at all (it runs in a vitual box). Any suggestions for why this might be?

The model information I am using is on the left of the figure, but please let me know if there is any other information I am missing. I'm a little perplexed! Thanks very much in advance.
Attachments
Screen Shot 2012-11-16 at 17.02.19.zip
Morphostats_figure1
(215.86 KiB) Downloaded 370 times
User avatar
riviere
Site Admin
Posts: 1361
Joined: Tue Jan 06, 2004 12:21 pm
Location: CEA NeuroSpin, Saint Aubin, France
Contact:

Re: Morphometric statistics problem

Post by riviere »

Hi Lewis,
Did you make sure you used labelled sulci graphs, for the same hemisphere as the model ?
Denis
lewishou
Posts: 8
Joined: Fri Nov 16, 2012 5:30 pm

Re: Morphometric statistics problem

Post by lewishou »

Yes, I think so, in that when I select the Data Type, I choose the Left Cortical Fold Graphs, and the model graph filename variable is Lfolds-model-fd4_native and the model is "folds_noroots_fd4_native_2010", with side on Left. Is this what you mean?
User avatar
riviere
Site Admin
Posts: 1361
Joined: Tue Jan 06, 2004 12:21 pm
Location: CEA NeuroSpin, Saint Aubin, France
Contact:

Re: Morphometric statistics problem

Post by riviere »

Yes but there are both labelled and unlabelled cortical folds graphs.
If you used unlabelled ones by mistake, then sulci are unnamed, and morphometric measurements for each sulcus are all 0 since sulci have not been identified.
Denis
lewishou
Posts: 8
Joined: Fri Nov 16, 2012 5:30 pm

Re: Morphometric statistics problem

Post by lewishou »

Ah I see...hopefully I haven't done that. Where can I check this? I assume this mistake would have been made in Morphologist at the processing level? Thanks, Lewis
User avatar
riviere
Site Admin
Posts: 1361
Joined: Tue Jan 06, 2004 12:21 pm
Location: CEA NeuroSpin, Saint Aubin, France
Contact:

Re: Morphometric statistics problem

Post by riviere »

The sulci recognition is the last step of the Morphologist pipeline. It is not done by default because it is a bit long and not everybody makes use of it. And moreover the sulci recognition models are not provided in the main package because of their size (but they are installed when needed).
Denis
lewishou
Posts: 8
Joined: Fri Nov 16, 2012 5:30 pm

Re: Morphometric statistics problem

Post by lewishou »

I assume you're talking about the SPAM models. Perhaps this is the problem, but I don't know how I could have changed the settings into using unlabelled cortical fold graphs as I have SPAM installed like last time, and the spam method is Global and Local. Have I simply installed the wrong ones, or chosen a stray wrong option somewhere? The SPAM checks say they are present and correct. Thanks.
User avatar
riviere
Site Admin
Posts: 1361
Joined: Tue Jan 06, 2004 12:21 pm
Location: CEA NeuroSpin, Saint Aubin, France
Contact:

Re: Morphometric statistics problem

Post by riviere »

Or maybe you just forgot to check the recognition step, since it is not checked by default ? Or you didn't pay attention to fill in the morphometry process with the labelled sulci ? Anyway, did you get the right figures now ?
Denis
lewishou
Posts: 8
Joined: Fri Nov 16, 2012 5:30 pm

Re: Morphometric statistics problem

Post by lewishou »

Ah yes, it was a problem with the labelled and unlabelled cortical graphs - I was using the left/right graphs rather than labelled, like you initially suggested. Apologies for my misunderstanding, thank you again.
Post Reply