Hi John, Those tweaks were very helpful. I looked at the volumes with the surface overlaid, and here's what I found:
http://brainmap.wustl.edu/pub/donna/IN/JOHN/align.html login pub password download I don't trust Caret to handle oblique volumes properly. Quoting the link above, "Can you go back through the history of this volume's processing -- specifically what happened to generate anatomical_ac_centered.hdr from anatomical_image.hdr? Were there steps the de-obliqued, flipped, or shifted the anatomical to get it AC-centered and probably in "LPI" orientation, as Caret required to segment?" Donna On Aug 20, 2015, at 1:02 AM, [email protected] wrote: > Hello Donna, > I've made the necessary changes and uploaded the file. > > john. > > > > If you can find your *.params file, then don't worry about the find >> command, which was intended to help you locate it if you did not know its >> name/location already. >> >> I looked at your dataset, and you must do two things to help me help you: >> >> * Remove the spaces from the filenames. Replace them with _ characters. >> When I try to read, move, or otherwise manipulate these files, the spaces >> are misinterpreted by the Linux shell as separate arguments. >> >> * Add the *.params file. >> >> After you've made those changes, rename the directory john_renamed. Then >> zip it and upload it. I'll do my best to solve it. >> >> >> On Aug 19, 2015, at 1:23 AM, [email protected] wrote: >> >>>> Hello Donna, >>> >>> 1.I ve tried taking the XYZ min values from .PARAMS file and transformed >>> the overlay. This appears very subjective and error prone. >>> >>> 2. "Do this in your subject directory: >>>> >>>> find /my/subject/dir -name "*.params" |sort | xargs grep -i min >>> " >>> >>> I am not sure i understood this step properly >>> >>> >>> I am unable to coregister the functional image and anatomical image >>> properly. >>> I am sorry to trouble you again , but it would be great if you can take >>> a >>> look at the dataset again, which i have uploaded in a folder " data from >>> john.zip" at http://pulvinar.wustl.edu/cgi-bin/upload.cgi >>> >>> I doubt now that there is some issue within the procedure that we follow >>> in doing the analysis. So it would be best if you can check/reanalyse >>> the >>> dataset from very initial step itself. >>> >>> PS: >>> -anatomical image.hdr\img---unaltered structural T1 image. >>> -functional.hdr\img----basic SPM8 T2*image which is to be mapped >>> >>> Thank you. >>> >>> >>> >>> >>> On Aug 18, 2015, at 2:19 AM, [email protected] wrote: >>>> >>>>> Hello Donna, >>>>> Thank you for your reply. >>>>> Two doubts i have >>>>> 1. why even after loading metric as primary overlay it is not getting >>>>> 'selectable' here in functional view (see attachment "capture")? >>>> >>>> The metric is a vertex-intensity mapping. It is not the volume. You >>>> can >>>> load the volume that was mapped using File: Open Data File: Volume >>>> Functional files. Then it will be selectable when you map to loaded >>>> volume. Or you can simply map to file on disk without loading. But it >>>> is >>>> not a bad idea to load the volume, too, to make sure everything aligns >>>> properly: Functional with surface is the important one, but the >>>> anatomical volume is the link between functional and surface (i.e., how >>>> they get aligned). >>>> >>>>> 2. what is the meaning of this error message (attachment 2), which >>>>> appears >>>>> on selecting the functional volumes? >>>> >>>> Again, the funky file naming of two of the volume files (e.g., space, >>>> parentheses, leading dashes) impedes my ability to check them quickly. >>>> But the whole brain anatomical does appear to be a NIfTI volume, rather >>>> than just an Analyze .hdr file. I loaded it as a functional volume, >>>> and >>>> then tried to map it to your surface. I got the same result as trying >>>> to >>>> map it from disk (clicking OK on the stickup you captured). >>>> >>>> That warning never got removed after support for nifti .hdr/.img pairs >>>> was >>>> added, but based on my getting the same results using the two paths >>>> mentioned above, I think you will solve your problem when you solve the >>>> misalignment between your cropped volume and the whole brain anatomical >>>> volume. Alternatively, shift the surface to meet the whole brain / >>>> functional volume. >>>> >>>> Ideally, get the following loaded and aligned in caret: >>>> >>>> * whole brain anatomy volume >>>> * functional volume overlay >>>> * surface (probably shifted version of what you have now) >>>> >>>> Do this in your subject directory: >>>> >>>> find /my/subject/dir -name "*.params" |sort | xargs grep -i min >>>> >>>> Capture it to a file if it's a lot. One of those files has the offset >>>> you >>>> need. >>>> >>>>> thank you. >>>>> >>>>> >>>>> Hi John, >>>>>> >>>>>> Got your upload. While I couldn't open the cropped volume in caret >>>>>> due >>>>>> to >>>>>> the way it was named, I was able to view the surface contour over the >>>>>> uncropped volume. See attached capture, which shows an offset. >>>>>> >>>>>> If you have a SureFit/Caret .params file (not included in the zip), >>>>>> it >>>>>> might contain the [XYZ]min values from the cropping, which might be >>>>>> used >>>>>> to either adjust the functional volume's origin, or more likely apply >>>>>> an >>>>>> affine transform to the surface, to shift it back into alignment with >>>>>> the >>>>>> whole brain volume. You don't have to blow away your existing coord; >>>>>> just >>>>>> rename the shifted version to indicate the offset. This can be done >>>>>> via >>>>>> command line or using the Caret: Window: Transformation Matrix >>>>>> editor. >>>>>> The polarity of the shift (+ or -) depends on whether you're shifting >>>>>> the >>>>>> volume or surface, and I always get confused about it. Usually I try >>>>>> it >>>>>> one way; look at the result like the capture below; and if it looks >>>>>> worse, >>>>>> I try it the other way. ;-) One of the ways usually does the trick. >>>>>> >>>>>> Donna >>>>>> >>>>>> >>>>>> >>>>>> On Aug 12, 2015, at 9:14 AM, Donna Dierker <[email protected]> >>>>>> wrote: >>>>>> >>>>>>> Could you upload your dataset in a zip archive here: >>>>>>> >>>>>> >>>>>>> http://pulvinar.wustl.edu/cgi-bin/upload.cgi >>>>>>> >>>>>>> Specifically I need: >>>>>>> >>>>>>> * functional volume being mapped >>>>>>> * anatomical volume with which functional volume aligns >>>>>>> * anatomical volume used to generate segmentation -- *cropped* >>>>>>> * fiducial coord file >>>>>>> * topology file >>>>>>> >>>>>>> I am wondering if the centers of the volumes between the cropped >>>>>>> volume >>>>>>> used to generate the surface and the whole brain anatomical volume. >>>>>>> >>>>>>> >>>>>>> On Aug 12, 2015, at 1:11 AM, [email protected] wrote: >>>>>>> >>>>>>>>> >>>>>>>> >>>>>>>> Ya, sorry for the incomplete query. >>>>>>>> Our interest is to view and threshold fMRI/voxel correlation data >>>>>>>> over >>>>>>>> the >>>>>>>> fiducial brain surface created out of high res image of individual >>>>>>>> macaques. >>>>>>>> 1. we used CARET 5.65 for creating the fiducial surfaces(following >>>>>>>> the >>>>>>>> caret5 tutorial guide for segmentation).we used individual already >>>>>>>> coregistered high res images for this purpose. >>>>>>>> 2. then we tried overlaying the fMRI data(spmT file from analysis >>>>>>>> using >>>>>>>> spm8) on the fiducial surface (following the procedures from >>>>>>>> http://prefrontal.org/blog/2009/04/using-caret-for-fmri-visualization/ >>>>>>>> and the tutorial guides). >>>>>>>> >>>>>>>> we ended up having >>>>>>>> a) fiducial surfaces with fmri data mapped onto it. But the voxel >>>>>>>> coordinates did not match properly with the results in spm8. >>>>>>>> I should tell you one thing that we found: when we overlay whole >>>>>>>> brain >>>>>>>> fmri results, it matches more or less in x&y axes but not in z >>>>>>>> axis. >>>>>>>> and >>>>>>>> when we overlay results from explicitly masked analysis(roi), it >>>>>>>> seems >>>>>>>> to >>>>>>>> be shifted caudally by 2-3mm in R-C axis. same things when checked >>>>>>>> in >>>>>>>> spm8, shows up to be well coregistered! >>>>>>>> >>>>>>>> b)Another issue is we are not able to threshold the caret brain >>>>>>>> overlays >>>>>>>> in concordance with the thresholds that we use in spm. >>>>>>>> The metric thresholding range (user defined) doesn't seems to >>>>>>>> represent >>>>>>>> what we really need. >>>>>>>> eg:we tried overlaying FDR corrected voel corrln values, thresh 0.4 >>>>>>>> on >>>>>>>> the >>>>>>>> brain in caet and tried a range of metric thresholds(user scale >>>>>>>> 0-1, >>>>>>>> put >>>>>>>> display mode-pos,thresh type -user, tried changing user pos thresh >>>>>>>> values >>>>>>>> to 0.4 and voxels doesn't show up! >>>>>>>> >>>>>>>> Thank you donna, >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> What software was used to reconstruct the surface? >>>>>>>>> >>>>>>>>> With freesurfer, there is an offset between the orig.mgz and the >>>>>>>>> surface. >>>>>>>>> And depending on many factors, you might have to flip/rotate the >>>>>>>>> surface >>>>>>>>> to be in the same orientation as the volume (or bring the volume >>>>>>>>> to >>>>>>>>> the >>>>>>>>> surface). >>>>>>>>> >>>>>>>>> See this thread: >>>>>>>>> >>>>>>>>> http://www.mail-archive.com/caret-users%40brainvis.wustl.edu/msg02081.html >>>>>>>>> >>>>>>>>> Also see the "Check Alignment between Normalized Volume and >>>>>>>>> Surface" >>>>>>>>> section here: >>>>>>>>> >>>>>>>>> http://brainvis.wustl.edu/help/pals_volume_normalization/spm5_normalization_pals.html >>>>>>>>> >>>>>>>>> Examining the surface contour overlaid on the volum in volume view >>>>>>>>> All >>>>>>>>> is >>>>>>>>> often very enlightening. >>>>>>>>> >>>>>>>>> > > _______________________________________________ > 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
