David and Donna, Thank you for this information. Yes, I'm planning on using SBR to an average surface of our monks (ala Argall et al). The atlas data appears to map onto both the F6 (registered to the F99) and F9 surfaces correctly. This seems correct since the F6 is registered to the F99. As David wrote, it shouldn't matter which surface I use. Regardless, I'll try SBR on both surfaces separately to our average surface, though I suspect any differences will be minimal.
I'm not sure what was going on when using a spec file that included both F6 and F99 surfaces, but it's not necessary for what I want to do, so I probably won't investigate further. The SBR should be straight forward from here. I just have to find my original surface files to register the atlas to. Thanks for these wonderful tools and the tech support! Mike On Oct 4, 2009, at 11:09 PM, David Van Essen wrote: > Mike, > > I hope your somewhat lengthy exchange with Donna successfully resolved > your main issues. But I didn't follow all of the gory details. Here, > I want to clarify (in case you and/or others don't know) what the > basic distinction is between F6 and F99. > > F99 is a single-subject rhesus macaque that has been our standard > atlas for years. The 'standard-mesh' left and right hem fiducial > surfaces, > Macaque.F99UA1.LEFT.FIDUCIAL.Std-MESH.73730.coord > Macaque.F99UA1.RIGHT.FIDUCIAL.Std-MESH.73730.coord > were registered to one another by landmark-based surface-based > registration (L-SBR), so that they are in correspondence - clicking on > a node in the left hem maps to the corresponding geographic (and > functional) location in the right hem. > > F6 is a population-average atlas (volume-registered) developed at Wash > U for fMRI studies, especially resting-state fMRI (Vincent et al., > 2007). The volume has very different dimensions than F99. > The standard-mesh left and right hem F6 surfaces, > Macaque.F6.LEFT.FIDUCIAL.Std-MESH.73730.coord > Macaque.F6.RIGHT.FIDUCIAL.Std-MESH.73730.coord > were registered to F99 by L-SBR. Hence, surface nodes of F6 and F99 > (either side) are in correspondence despite the difference in spatial > dimensions. > > If you are planning to make comparisons with other macaque datasets > using SBR, then it won't matter a whole lot whether you use F99 or > F6. However, I would recommend F99 because far more data have been > mapped directly to F99 by one or another method, so it is in some > respects more 'primary'. > > Finally, as an aside for those who care about monkey-human > comparisons, it is important to know that what is 'standard-mesh' for > macaque doesn't make it in functional (or geographic) correspondence > with the 'standard-mesh' PALS atlas. Both sets of surfaces are 73,730 > nodes, but they are essentially independent of one another. We have > published on using SBR to achieve interspecies registration (Van Essen > and Dierker, 2007), and it leads to a different set of coordinate > files and deformation maps that are a story unto themselves. > > I hope this is helpful, at least for somebody. > > David VE > > On Oct 2, 2009, at 11:35 AM, Donna Dierker wrote: > >> My attempt to clarify made it even more confusing: 'I think you >> mean "could maintain node correspondence when switching >> between F6 and F99 surfaces in the same .spec file," based on my >> understanding of your previous paragraph.' >> >> You meant could NOT maintain node correspondence. >> >> >> On 10/02/2009 09:37 AM, Donna Dierker wrote: >>> See inline below. >>> >>> On 10/01/2009 02:34 PM, Michael Arcaro wrote: >>> >>>> Hi Donna, >>>> >>>> Thanks for the heads up. I'm was a bit confused as to the >>>> differences >>>> between Caret and Surefit datasets. I tried using SUMA's Make_Spec >>>> script for Caret surfaces. I noticed that I had to rename the files >>>> slightly for the program to recognize the .coord and .topo files. >>>> >>>> I also had to edit the SUMA .spec file. When, I initially included >>>> both F6 and F99 datasets, the node correspondence between surfaces >>>> was >>>> lost. By creating .spec files that only load F6 or F99 surfaces, >>>> node >>>> correspondence is maintained when switching between surfaces >>>> (within >>>> F6 or F99 datasets) in SUMA. >>>> >>>> >>> Hmmm. No clue there, but I bet Ziad would have some ideas. >>> >>>> The next issue was mapping on the curvature. As you suggested, >>>> extracting the curvature column and making a 1D.dset did the trick. >>>> Interestingly, this 1D.dset file works for both F6 and F99 datasets >>>> and appears to have proper mapping for surfaces in both datasets. >>>> This >>>> seems a bit odd since I could maintain node correspondence when >>>> switching between F6 and F99 surfaces in the same .spec file. >>>> >>>> >>> I think you mean "could maintain node correspondence when switching >>> between F6 and F99 surfaces in the same .spec file," based on my >>> understanding of your previous paragraph. >>> >>> Again, I can't explain what you saw in the paragraph preceding my >>> "Hmmm" >>> response above. >>> >>> But I can explain why the 1D works for both: David took great pains >>> to >>> register both F6 and F99 to the standard 73730 mesh. Thank David >>> Van Essen. >>> >>>> I also extracted the border projections into a .1D.dset which maps >>>> onto the surfaces fine. It's a bit of a hack job, but I haven't >>>> found >>>> any tutorials on SUMA on how to properly load .paint or .borderproj >>>> files. This also seems to properly map onto both F6 and F99 >>>> datasets. >>>> >>>> >>> This is a first to my knowledge. I'm not sure how SUMA handles >>> borders. >>> >>>> This is what I really care about. I hope to use the SUMA tools to >>>> warp >>>> the F6 dataset and my own datasets into a standard / aligned space. >>>> >>>> >>> Are you using something like the Argall, et. al. method to register >>> your >>> subjects? >>> >>>> All my data is on the surface. I don't think I will need to project >>>> back into volume space, so the point of origin should not be an >>>> issue >>>> for me. >>>> >>>> Mike >>>> >>>> >>>> On Oct 1, 2009, at 2:24 PM, Donna Dierker wrote: >>>> >>>> >>>> >>>>> Mike, >>>>> >>>>> Just as a heads up, when SUMA was first developed, Caret and >>>>> SureFit >>>>> were separate. (SureFit generated segmentations and surfaces; >>>>> Caret >>>>> displayed, flattened, and registered surfaces.) In those days, >>>>> SureFit's origin was at the lower left corner of the cropped >>>>> bounding >>>>> box. When SureFit merged into Caret, there was a migration from >>>>> voxel >>>>> space to coordinate space, and the origin moved to the AC (or >>>>> whatever >>>>> the origin in the volume fed to Caret for segmentation). If SUMA >>>>> translates the surface by some parameters defined in the .params >>>>> file, >>>>> you'll see major misalignment between surface and volume. I think >>>>> Ziad >>>>> has a work-around for this -- perhaps a flag added to the coord >>>>> header >>>>> to indicate it is post-SureFit. The F6 and F99 datasets are post- >>>>> SureFit. >>>>> >>>>> It is possible that you'll need to convert the surface shape file >>>>> to >>>>> ASCII ; strip the header; and extract the curvature column into a >>>>> "1D" >>>>> file. But it is worth posting to the AFNI message board. >>>>> >>>>> Donna >>>>> >>>>> On 10/01/2009 12:34 PM, Michael Arcaro wrote: >>>>> >>>>> >>>>>> Thanks Donna, >>>>>> >>>>>> I'll use the F6.BOTH.withF99 dataset. I've had success loading >>>>>> individual surfaces into SUMA (after converting the files to >>>>>> ASCII), >>>>>> but haven't figured out how to load / display the curvature. >>>>>> Since >>>>>> this is probably more related to SUMA and figuring out how to >>>>>> properly >>>>>> load the data with SUMA, I'll follow up with the AFNI/SUMA folk >>>>>> for >>>>>> more help. >>>>>> >>>>>> It also appears that SUMA is looking for a .params file. I >>>>>> generated >>>>>> one in Caret from File-> save data file (after loading the atlas >>>>>> spec). I assume this should be sufficient. >>>>>> >>>>>> Thanks! >>>>>> Mike >>>>>> >>>>>> >>>>>> On Oct 1, 2009, at 9:38 AM, Donna Dierker wrote: >>>>>> >>>>>> >>>>>> >>>>>> >>>>>>> Hi Mike, >>>>>>> >>>>>>> 1) F6 dataset: Probably only David knows for sure, but >>>>>>> Macaque.F6.BOTH.SURF-VOL.Std-MESH.73730 is dated 3/9/2007, while >>>>>>> Macaque.F6.BOTH.withF99-Data.Std-MESH.73730 is dated 8/16/2007. >>>>>>> Since >>>>>>> they have many files in common, my guess is that the latter was >>>>>>> based on >>>>>>> the former, but has an important enhancement: >>>>>>> Macaque.F99UA1.COMPOSITE_PartititioningSchemes. >>>>>>> 73730.borderproj. It >>>>>>> has >>>>>>> the partitioning schemes not only in paint form, but also border >>>>>>> form. >>>>>>> I'd use the latter. >>>>>>> >>>>>>> 2) SUMA interoperability: While John Harwell (our lab) has >>>>>>> worked >>>>>>> with >>>>>>> Ziad Saad and Rick Reynolds (AFNI/SUMA orbit) on GIFTI, a >>>>>>> neutral >>>>>>> surface file format akin to what NIfTI does for volumes, I don't >>>>>>> think >>>>>>> you need to convert your surfaces to GIFTI in order for SUMA to >>>>>>> read >>>>>>> them. Ziad has worked pretty hard to make SUMA read Caret >>>>>>> surfaces >>>>>>> with >>>>>>> little tweaking. We did uncover some issues recently with the >>>>>>> way >>>>>>> Caret >>>>>>> was writing AFNI .HEAD files, but sticking with NIfTI, where >>>>>>> possible, >>>>>>> should avoid these problems. That said, both the datasets >>>>>>> mentioned >>>>>>> above have AFNI volumes, so if you run into any alignment issues >>>>>>> between >>>>>>> the volume and the surface in SUMA, let us know. I don't >>>>>>> necessarily >>>>>>> anticipate problems. >>>>>>> >>>>>>> Donna >>>>>>> >>>>>>> On 09/30/2009 03:15 PM, Michael Arcaro wrote: >>>>>>> >>>>>>> >>>>>>> >>>>>>>> Hello, >>>>>>>> >>>>>>>> I'm interested in mapping some macaque fMRI data onto the >>>>>>>> macaque >>>>>>>> atlases, specifically the Lewis - Van Essen '00 atlas. I'm new >>>>>>>> to >>>>>>>> Caret and have a couple of basic questions: >>>>>>>> >>>>>>>> First, is there a recommended atlas dataset to use? (F6 or F99) >>>>>>>> From >>>>>>>> the tutorial, it sounds as if I would want to go with the F6 >>>>>>>> atlas. >>>>>>>> Assuming I go with the F6, what are the differences between: >>>>>>>> Macaque.F6.BOTH.SURF-VOL.Std-MESH.73730 >>>>>>>> Macaque.F6.BOTH.withF99-Data.Std-MESH.73730 >>>>>>>> Is there a reason to use one of these above the other? >>>>>>>> >>>>>>>> Second, I'm looking to import the atlas into SUMA to warp/align >>>>>>>> with >>>>>>>> my data. I'm having a bit of difficulty with this. From what I >>>>>>>> have >>>>>>>> read, SUMA can read Caret format. I was wondering if I need to >>>>>>>> convert >>>>>>>> any of the files into a different format for SUMA to read (or >>>>>>>> possibly >>>>>>>> rename the files). This may be more of a question for the SUMA >>>>>>>> folk, >>>>>>>> but I thought I'd see if anyone here is familiar with >>>>>>>> importing the >>>>>>>> atlases into SUMA. >>>>>>>> >>>>>>>> Thanks, >>>>>>>> Mike Arcaro >>>>>>>> >>>>>>>> >>>>>>>> >>>>> _______________________________________________ >>>>> 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 >>>> >>>> >>> >>> _______________________________________________ >>> 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 > > _______________________________________________ > 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
