Honestly, I think I would just map your fMRI and VBM results on the PALS-B12 SPM5 average fiducial surface, because short of segmenting all your subjects; generating their surfaces; registering them to some atlas target; and generating an average fiducial surface of your subjects, I don't think the alternatives below will be any better. They could easily be worse.
On 10/28/2010 11:46 AM, Iroise Dumontheil wrote: > Hi, > > I am new to Caret and would like to use it to plot results from fMRI and VBM > analyses from a single study on a mean cortical surface brain from our > participants. I used Dartel and SPM5 for the VBM analyses and I have a mean > grey matter image from our group of participants (as .hdr + .img) (dimensions > 121 x 145 x 121, datatype int16, values when plotted between 0 and 1.0722, > voxel size 1.63 x 1.56 x 1.68). > > I have tried to draw contours (white to grey border) from coronal images > generated using MRIcro to make a hemisphere from this data. It kind of > worked, although I am not sure how to best inflate this surface so that it > corresponds to the cortical surface rather than the white to grey border (at > the moment some fMRI activations do not get mapped on the surface because > there are too far from it in a grey matter thick area). > You can use Surface: Geometry: Expand surface, but I've never done this to approximate midthickness from WM/GM boundary -- at least not for publication purposes. You can try using MCW Brainfish mapping algorithm to see if it maps a bit better, but there are limits to what it can do. > My question is whether there is any using Caret that I could generate the > cortical surface automatically from my mean image? The mean image is quite > neat (I could send it if necessary). I have seen options to segment > structural images, but I am wondering whether there is a way to step in the > process at a later stage with this mean grey matter image. > You can try sending me the mean image: http://pulvinar.wustl.edu/cgi-bin/upload.cgi Every time I've ever tried to do something like this, it hasn't worked. I realize mean images are getting crisper and crisper as volume registration techniques improve; however; there is still patching, etc., that will need doing, and you won't have all the PALS goodies unless you register it to that atlas. Do this: At least try mapping both your fMRI and VBM results to the PALS atlas using the procedures on page 67 of this tutorial: http://brainvis.wustl.edu/wiki_linked_files/documentation/Caret_Tutorial_Sep22.pdf Use the SPM5 space surfaces. Have a look at those results, and decide whether it's worth segmenting a mean image. > Thanks for your help! > > Iroise > _______________________________________________ > caret-users mailing list > [email protected] > http://brainvis.wustl.edu/mailman/listinfo/caret-users > _______________________________________________ caret-users mailing list [email protected] http://brainvis.wustl.edu/mailman/listinfo/caret-users
