>> Subject: Sections and MRIs >> From: "Peter L. Carras" <pcarr...@pitt.edu> >> Date: Fri, 26 Jun 2009 18:43:04 -0400 >> To: caret-users@brainvis.wustl.edu >> >> To: caret-users@brainvis.wustl.edu >> >> >> Hello, >> >> I recently started using CARET (v5.61). I have had some success, but >> also a few questions that do not seem to be covered in the tutorials. >> We have macaque brains with both MRI images and histological >> sections. The layer four contours for the sections of a left frontal >> cortex were traced in MDPlot. I have successfully imported the >> contours into CARET, along with the labeled neurons found in each >> section, and saved them as a spec file. I also opened the MRI image >> of the whole head in CARET, segmented it, and saved the left cortical >> surface as another spec file. >> >> Q 1 - Is it possible to display the labeled neurons from the section >> data on the MRI image of the complete cortex? If so, how would I go >> about doing this? The histological sections were cut in the coronal >> plane, and extend caudally only to the posterior commissure.
Not having taking the tutorial on this in years, my guess is that the 'labeled neurons' are stored as "cell data" (aka stereotaxic foci). If so, then foci can be displayed not only on surfaces, but also on volumes. MRI and sections/cells must be in register -- probably using a process outside of Caret. >> Q 2 - When the contours were originally drawn, successive sections >> were aligned on the dorsal-medial boundary of each section. The >> cortical fiducial surface as reconstructed by CARET is therefore >> somewhat distorted - the sulci are visible but the dorsal edge of the >> medial wall is flat rather than curved. Will this cause the labeled >> neurons to appear in incorrect locations when they are displayed on >> the MRI-based surface? I'm guessing not, if I'm right about them being stored as cell data. Again, the tricky bit is getting sections/cells in volumetric register with the MRI. Note here the distinction between a .cell or .foci file and their corresponding .cellproj or .fociproj. The *proj files are projected to surface tiles, so errors in the surface will propagate to the projected foci. But cells/foci in .cell or .foci files aren't subject to the surface's baggage. If you want to view the cells/foci on alternate configurations like inflated and flat, then projection is how you do that. >>If so, can you suggest a way to correct the surface (short of re-tracing all 390 sections)? I'd have to see the surface to give an opinion about how to fix it. >> Q 3 - The SureFit segmentation algorithms were unable to successfully >> operate on the MRI image, even when I tried varying the default gray >> matter and white matter peak values. Typically, the program stopped >> and reported an error in the eye fat segmentation step, or an error in >> hindbrain segmentation. When I de-selected those steps, CARET crashed >> when determining layer 4 or filling the ventricles. I ended up >> segmenting most of the brain manually. Do you have advice on how to >> increase the likelihood that SureFit will succeed in segmenting a >> macaque MRI image? The MPRAGE scan was done on a 3T Siemens Allegra >> scanner with 0.7 mm isotropic voxels, and was rotated in CARET to LPI >> orientation. Did you do any non-uniformity correction? The Allegra 3T is not my top choice for anatomical data, but if you compare its MPRAGE to what we used to see with surface coils, I realize I'm spoiled now. It might be worth trying FSL's fast to bias correct the image, and use the "*restore*" version of the volume. Typically you must BET before FAST, so turning off eye removal will make sense if you've skull-stripped the brain with BET. I don't think turning off hindbrain works, so leave that on. If bias correction fails to get you a segmentation, then I'll ask you to upload the volume so I can have a look Monday. You did crop it to left or right hemisphere, right? >> Q 4 - Is it possible within CARET to convert the map of labeled >> neurons (shown as contour cells on the fiducial surface) to a map of >> cell density? I don't know. Have you scanned the tutorials on this? >> Q 5 - In addition to the data on macaques, we also have anatomical and >> MRI data from new world monkeys (Cebus apella). Do you anticipate >> adding Cebus to the list of species in the Metadata category? I'll leave this one for David. >> >> >> Your advice or assistance would be much appreciated. >> >> Sincerely, >> Peter Carras >> > > -- > Peter L. Carras, Ph.D. > Systems Neuroscience Institute > University of Pittsburgh > > _______________________________________________ > caret-users mailing list > caret-users@brainvis.wustl.edu > http://brainvis.wustl.edu/mailman/listinfo/caret-users > _______________________________________________ caret-users mailing list caret-users@brainvis.wustl.edu http://brainvis.wustl.edu/mailman/listinfo/caret-users
