[caret-users] Problem viewing NIFTI volume file in 5.6
Baylis Shanks
bshanks3 at hotmail.com
Fri Oct 10 21:10:02 CDT 2008
thanks,
i tried scaling my coord file as you suggested. it works, but the result of the mapping is "patchier" than it was with the old Caret, at least with the ENCLOSING_VOXEL algorithm.
Here's a metric file generated using the old v5.51 and an unscaled coordinate file:
http://rapidshare.com/files/152840996/test_old_caret.metric.html
http://rapidshare.com/files/152841538/test.coord.html
And here's a metric file generated using v5.601 and a scaled coordinate file:
http://rapidshare.com/files/152840997/test_scaled.metric.html
http://rapidshare.com/files/152840995/test_scaled.coord.html
Using the Gaussian algorithm I can get it to smudge it enough to fill in the empty nodes, but i'm wondering if the patchiness with ENCLOSING_VOXEL is a sign of a problem?
The patchiness is still there even if i regenerate the surface from scratch using v5.601.
thx, b
----------------------------------------
> From: john at brainvis.wustl.edu
> To: caret-users at brainvis.wustl.edu
> Date: Thu, 9 Oct 2008 10:13:12 -0500
> Subject: Re: [caret-users] Problem viewing NIFTI volume file in 5.6
>
> Baylis,
>
> Previously, you discovered that Caret was not reading a NIFTI volume
> correctly in regards to the units and if only the qform coordinate
> space was used. When you generated your surface, it must have been
> generated with a version of Caret that did not correctly read the
> volume as the surface's units are microns. If you download updated
> versions of care5 and caret_command it should resolve your problem.
> You will need to either regenerate your surface, so that its units is
> millimeters, or scale the surface using Surface Menu->Transform->Scale
> with a value of 0.001 and save the coordinate file.
>
> To verify that the surface and the volume are properly aligned, load
> both the volume and the surface into Caret and display the volume.
> Use the Display Control's Volume Surface Outline page to place an
> outline of the surface over the volume.
>
> Updated versions of caret5 (caret5_exe_*.zip) and caret_command
> (caret_command_exe_*.zip) are available at http://brainmap.wustl.edu/pub/john/
> . Use the username "pub" and the password "download" to access the
> website. Source code (caret_source.v5.6.zip) is also available.
>
> -----------------------------------
> John Harwell
> john at brainvis.wustl.edu
>
> Department of Anatomy and Neurobiology
> Washington University School of Medicine
> 660 S. Euclid Ave Box 8108
> Saint Louis, MO 63110
>
>
>
>
> On Oct 8, 2008, at 9:19 PM, Baylis Shanks wrote:
>
>>
>> hello again,
>>
>>
>> When using caret_command to do a -volume-map-to-surface i seem to be
>> getting a bunch of zeros in the output metric file with at least
>> some mapping algorithms.
>>
>> If i use caret_map_fmri version 5.51 I don't get zeros.
>>
>> Specifically, the command that produces non-zeros is:
>>
>> caret_map_fmri -a EV -v test.nii -s test.spec test.topo test.coord -
>> m test.metric
>>
>> and the command that produces zeros is:
>>
>> caret_command -volume-map-to-surface test.coord test.topo \"\"
>> test.metric METRIC_ENCLOSING_VOXEL test.nii
>>
>> note that the .coord and .topo files that i am using in both cases
>> were produced by Caret 5.51.
>>
>>
>> test.spec, test.topo, and test.coord are available at:
>>
>> http://rapidshare.com/files/152231400/test.spec.html
>> http://rapidshare.com/files/152231401/test.topo.html
>> http://rapidshare.com/files/152231402/test.coord.html
>>
>> test.nii is the same as before.
>>
>> in case it's useful, in attempting (and failing) to debug the
>> problem, i observed what may be inappropriate units either being fed
>> to or generated by the function convertCoordinatesToVoxelIJK.
>> details below
>>
>>
>> -----
>>
>>
>>
>> in debugging the problem, i noticed that in the file
>> BrainModelVolumeToSurfaceMapper.cxx, in the function
>> getNeighborsSubVolume (which is called by some mapping algorithms),
>> there is a call to convertCoordinatesToVoxelIJK.
>> convertCoordinatesToVoxelIJK always returns FALSE in my case because
>> it thinks the coordinates it is given are out of bounds.
>>
>> debugging convertCoordinatesToVoxelIJK, i notice that the numbers
>> passed in as "xyz" seem to be in the thousands (i.e. numbers like
>> 4150.34 -- probably in units of microns) whereas the result of
>> getOriginAtCornerOfVoxel are in units of millimeters. the units of
>> "spacing" are also millimeters.
>>
>> If, inside of convertCoordinatesToVoxelIJK, i temporarily replace
>> the line:
>>
>> const float d = xyz[i] - originCorner[i];
>>
>> with
>>
>> const float d = (xyz[i]/1000) - (originCorner[i]);
>>
>> then the output of convertCoordinatesToVoxelIJK is more reasonable
>> (by which i mean, it doesn't return FALSE anymore). To me this
>> suggests that the function convertCoordinatesToVoxelIJK should
>> either be given its input in terms of millimeters, or should consult
>> the header information and convert its input units if needed.
>>
>> however, even with the temporary modification, the result of
>> caret_command -volume-map-to-surface is zeros, so there must be
>> something else also.
>>
>> thanks,
>> bayle
>>
>>
>>
>>
>>
>>
>> ----------------------------------------
>>> From: john at brainvis.wustl.edu
>>> To: caret-users at brainvis.wustl.edu
>>> Date: Tue, 23 Sep 2008 12:00:00 -0500
>>> Subject: Re: [caret-users] Problem viewing NIFTI volume file in 5.6
>>>
>>> Bayle,
>>>
>>> There are two ways to describe the coordinate system in a NIFTI
>>> volume
>>> by using the qform_code and the sform_code. Your volume used only
>>> the
>>> qform_code and Caret was processing it incorrectly. Some changes
>>> have
>>> been made and Caret now appears to read your volume correctly.
>>>
>>> You can get an updated version of the caret5 executable from "http://brainmap.wustl.edu/pub/john/
>>> ". Use the username "pub" and the password "download" to access the
>>> website. There are three zip files whose names begin with
>>> "caret5_exe" that contain updated versions of caret5. Download the
>>> version appropriate for your operating system and install it. On
>>> Windows, place caret5.exe in you Caret installations "bin_windows"
>>> directory. On Linux, place caret5 in you Caret installation's
>>> "bin_linux" directory. On Mac OSX, place caret5.app in you Caret
>>> installation's "macosx_apps" directory.
>>>
>>> -----------------------------------
>>> John Harwell
>>> john at brainvis.wustl.edu
>>>
>>> Department of Anatomy and Neurobiology
>>> Washington University School of Medicine
>>> 660 S. Euclid Ave Box 8108
>>> Saint Louis, MO 63110
>>>
>>>
>>>
>>>
>>> On Sep 23, 2008, at 8:27 AM, Donna Dierker wrote:
>>>
>>>> Hi Bayle,
>>>>
>>>> I removed your attachments and placed them in our upload directory,
>>>> and
>>>> I am forwarding your reply to caret-users that got held up due to
>>>> message size constraints. I let the first attachment you sent
>>>> through,
>>>> because it was just a bit above the limit, but this message had
>>>> three
>>>> volumes attached exceeding 2.5Mb total. We don't want to fill up
>>>> everyone's inbox.
>>>>
>>>> I'll try to have a look at this later today, but I wanted John
>>>> Harwell
>>>> to get a look before he takes a few days off later this week.
>>>>
>>>> Donna
>>>>
>>>> -------- Original Message --------
>>>> Subject: RE: [caret-users] Problem viewing NIFTI volume file in 5.6
>>>> Date: Tue, 23 Sep 2008 01:21:10 +0000
>>>> From: Baylis Shanks
>>>> To: Caret, SureFit, and SuMS software users
>>>>
>>>>
>>>>
>>>>
>>>> hi,
>>>>
>>>> thanks for looking into it.
>>>>
>>>>> When I view it in AFNI, I do see something that might be mouse
>>>>> cortex
>>>>
>>>> yes, it's mouse cortex, and that is what it is supposed to look
>>>> like :)
>>>>
>>>>
>>>>> R-to-L extent: 5688.600 [L] -to- 5700.000 [L] -step- 0.200 mm
>>>>> [ 58
>>>>> voxels]
>>>>> A-to-P extent: 7786.800 [P] -to- 7800.000 [P] -step- 0.200 mm
>>>>> [ 67
>>>>> voxels]
>>>>> I-to-S extent: -2700.000 [I] -to- -2692.000 [I] -step- 0.200 mm
>>>>> [ 41
>>>>> voxels]
>>>>> ...
>>>>> If the voxdims are really cubic .2mm, which is consistent with the
>>>>> mousy
>>>>> look of what AFNI shows, then the field of view should be 11.6x13.4
>>>>> mm
>>>>> -- not what is shown above.
>>>>
>>>> by "what is shown above", you meant this, right?
>>>>
>>>> there are indeed supposed to be 58, 67, and 41 voxels along the R-L,
>>>> A-P, and I-S directions. the value of the "dim" header field is
>>>> [ 3 58 67 41 0 0 0 0] (i think -- that is the
>>>> value that i tried to put in there, but bear in mind that since i
>>>> created the file i could have written out the header incorrectly).
>>>>
>>>> on to the displacement. i was trying to make the origin be the
>>>> anterior commissure, which i believe to be near 5.7, 7.8, 2.7 mm.
>>>>
>>>> what are the units of the numbers 5700, 7800, 2692? my intent was
>>>> for them to be microns. the values of the NIFTI header fields
>>>> qoffset_x, qoffset_y, qoffset_ z are (i think) -5700, -7800, and
>>>> -2700 respectively. the value of the NIFTI header field xyzt_units
>>>> is (i think) NIFTI_UNITS_MICRON. but perhaps the units of the
>>>> qoffset_* fields were supposed to be units of millimeters?
>>>>
>>>> 5700-5688.6 is 11.4, which is pretty close to 11.6 (58*.2 = 11.6) --
>>>> so it seems like 3dinfo is interpreting the 5700 as mm, and
>>>> subtracting 11.4 mm from it to get the 5688.6.
>>>>
>>>> so, i tried changing header fields but i haven't yet figured out how
>>>> to completely fix the problem (read on for details).
>>>>
>>>> before talking about that, i'd like to note something strange i
>>>> noticed in the Caret's Volume Attributes Editor when viewing the
>>>> original file that i included in the previous email, test.nii (when
>>>> loading it as an Anatomy volume). The dialog shows the "Dimensions"
>>>> as 1 in each direction, "Voxel Size" as 200, and Origin as -57, -78,
>>>> and -27. This is odd, because one would think that the Voxel Size is
>>>> correct and is displayed in units of microns, but one would think
>>>> that Origin would be either -5700, -7800, -2700 (if the qoffset
>>>> header fields are being read as microns and displayed as microns)
>>>> -5700000, -7800000, -2700000 (if the qoffset header fields are being
>>>> read as mm and displayed as microns), or it should be -5.7, -7.8,
>>>> -2.7 (if the qoffset header fields are being read as microns and
>>>> displayed as mm).
>>>>
>>>> now, i will describe the results of trying to change some of the
>>>> header fields, yielding new test files (attached).
>>>>
>>>> first, i tried to simply scale the qoffset fields. I replaced -5700,
>>>> -7800, -2700 with -5.7, -7.8, -2.7. The result is file test2_5.7.nii
>>>> (attached). Caret has the same problem as before. Caret's Volume
>>>> Attribute Editor shows "Dimensions" as 1 in each direction, "Voxel
>>>> Size" as 200, and Origin as -0.057, -0.078, and -0.027. Again, this
>>>> is odd; I would have expected one of 5700, -7800, -2700 (if the
>>>> qoffsett header fields are being read as mm and displayed as
>>>> microns) -5.7, -7.8, -2.7 (if the qoffset header fields are being
>>>> read as microns and displayed as microns), or it should be -0.0057,
>>>> -0.0078, -0.0027 (if the qoffset header fields are being read as
>>>> microns and displayed as mm).
>>>>
>>>> second, i tried setting the header field xyzt_units to
>>>> NIFTI_UNITS_MM. I changed the pixdim header field from [200 200 200]
>>>> to [.2 .2 .2], and i left the qoffset fields as -5.7, -7.8, -2.7
>>>> (test3_mm.nii, attached). Now Caret displays something, and it even
>>>> looks to be centered near the origin! However, the display looks too
>>>> small for some reason, and the Volume Attributes Editor shows
>>>> Dimensions as 286, 331, 201, Voxel Size as .2, .2, .2, and Origin as
>>>> -57, -78, -27. So, both the Dimensions and Origin field seems to be
>>>> wrong.
>>>>
>>>> third, i tried leaving xyzt_units as NIFTI_UNITS_MM, leaving the
>>>> pixdim header field as [.2 .2 .2], and rescaling the qoffset fields
>>>> by a factor of ten, to -.57, -.78, -.27, hoping to make the Origin
>>>> fields in Caret's Volume Attributes Editor come out right
>>>> (test4_mm_.
>>>> 57.nii). Loading this file into Caret, you can see something, but it
>>>> is really tiny, and not centered at the origin. The Volume
>>>> Attributes Editor shows Dimensions as 286, 331, 201, Voxel Size as .
>>>> 2, .2, .2, and Origin as -5.7, -7.8, -2.7. So, now if the units in
>>>> the Volume Attributes Editor are interpreted as mm, the Voxel Size
>>>> and Origin are both correct, however the Dimensions are still wrong.
>>>>
>>>> thanks,
>>>> bayle
>>>>
>>>>
>>>> ----------------------------------------
>>>>> Date: Fri, 19 Sep 2008 08:33:33 -0500
>>>>> From: donna at brainvis.wustl.edu
>>>>> To: caret-users at brainvis.wustl.edu
>>>>> Subject: Re: [caret-users] Problem viewing NIFTI volume file in 5.6
>>>>>
>>>>> Hi Bayle,
>>>>>
>>>>> Your problem replicates on my end -- i.e., I get an empty volume
>>>>> when I
>>>>> view test.nii in Caret (see test.nii.caret.jpg).
>>>>>
>>>>> When I view it in AFNI, I do see something that might be mouse
>>>>> cortex
>>>>> (see test.nii.afni.jpg).
>>>>>
>>>>> The header does look unusual. Here is AFNI's 3dinfo output:
>>>>>
>>>>> it.wustl.edu 30% 3dinfo test.nii
>>>>> ++ 3dinfo: AFNI version=AFNI_2007_05_29_1644 (Jan 29 2008) [64-bit]
>>>>>
>>>>> Dataset File: test.nii
>>>>> Identifier Code: NII_-z--ofl-dASZaHy2pulrNg Creation Date: Fri Sep
>>>>> 19
>>>>> 08:26:32 2008
>>>>> Dataset Type: Anat Bucket (-abuc)
>>>>> Byte Order: LSB_FIRST {assumed} [this CPU native = LSB_FIRST]
>>>>> Storage Mode: NIFTI file
>>>>> Storage Space: 637304 (637 thousand) bytes
>>>>> Data Axes Tilt: Plumb
>>>>> Data Axes Orientation:
>>>>> first (x) = Left-to-Right
>>>>> second (y) = Posterior-to-Anterior
>>>>> third (z) = Inferior-to-Superior [-orient LPI]
>>>>> R-to-L extent: 5688.600 [L] -to- 5700.000 [L] -step- 0.200 mm
>>>>> [ 58
>>>>> voxels]
>>>>> A-to-P extent: 7786.800 [P] -to- 7800.000 [P] -step- 0.200 mm
>>>>> [ 67
>>>>> voxels]
>>>>> I-to-S extent: -2700.000 [I] -to- -2692.000 [I] -step- 0.200 mm
>>>>> [ 41
>>>>> voxels]
>>>>> Number of values stored at each pixel = 1
>>>>> -- At sub-brick #0 '?' datum type is float
>>>>>
>>>>> If the voxdims are really cubic .2mm, which is consistent with the
>>>>> mousy
>>>>> look of what AFNI shows, then the field of view should be 11.6x13.4
>>>>> mm
>>>>> -- not what is shown above.
>>>>>
>>>>> Let's start by verifying the voxdims and field of view.
>>>>>
>>>>> Donna
>>>>>
>>>>> On 09/18/2008 05:17 PM, Bayle Shanks wrote:
>>>>>> Hi, I'm having trouble viewing the attached NIFTI functional
>>>>>> volume
>>>>>> file (test.nii) in Caret 5.56. I can view it fine in Caret 5.51.
>>>>>>
>>>>>> What happens is that when i open the file, there are no errors but
>>>>>> it
>>>>>> looks empty. The three coordinate selectors at the top (to the
>>>>>> left of
>>>>>> the D/C button) are each at 0 and can't be changed.
>>>>>>
>>>>>> I created the file
>>>>>> myself so it's possible that it doesn't conform to the NIFTI
>>>>>> standard
>>>>>> in some way (of course, if so i'd like to correct that).
>>>>>>
>>>>>> I'm using Debian 2.6.18-6-amd64 (64-bit) (a mix of Debian stable
>>>>>> and
>>>>>> unstable).
>>>>>>
>>>>>> I've tried building Caret from source and also running the binary
>>>>>> distribution. I've tried caret5 and also caret5_mesa. With
>>>>>> caret5_mesa, some errors are displayed on the console (not sure if
>>>>>> I was
>>>>>> using CARET_DEBUG), I've copied them below.
>>>>>>
>>>>>> thanks,
>>>>>> bayle
>>>>>>
>>>>>> -----
>>>>>> Time to read .caret5_preferences (0.00209236 MB) was 0 seconds.
>>>>>> Caret Home Directory: /usr/local/caret
>>>>>> NIFTI extension[0] 0
>>>>>> Unscaled range: 0 10.9652
>>>>>> Scaled range: 0 10.9652
>>>>>>
>>>>>> OpenGL Error: invalid enum
>>>>>> At end of drawBrainModelPrivate().
>>>>>> While drawing brain model VOLUME - test.nii
>>>>>> In window number 0
>>>>>> Projection Matrix Stack Depth 1
>>>>>> Model Matrix Stack Depth 1
>>>>>> Name Matrix Stack Depth 0
>>>>>>
>>>>>
>>>>
>>>> _________________________________________________________________
>>>> See how Windows Mobile brings your life together—at home, work, or
>>>> on the go.
>>>> http://clk.atdmt.com/MRT/go/msnnkwxp1020093182mrt/direct/01/
>>>>
>>>> _______________________________________________
>>>> caret-users mailing list
>>>> caret-users at brainvis.wustl.edu
>>>> http://brainvis.wustl.edu/mailman/listinfo/caret-users
>>>>
>>>
>>>
>>> _______________________________________________
>>> caret-users mailing list
>>> caret-users at brainvis.wustl.edu
>>> http://brainvis.wustl.edu/mailman/listinfo/caret-users
>>
>> _________________________________________________________________
>> Want to do more with Windows Live? Learn “10 hidden secrets” from
>> Jamie.
>> http://windowslive.com/connect/post/jamiethomson.spaces.live.com-Blog-cns!550F681DAD532637!5295.entry?ocid=TXT_TAGLM_WL_domore_092008
>> _______________________________________________
>> caret-users mailing list
>> caret-users at brainvis.wustl.edu
>> http://brainvis.wustl.edu/mailman/listinfo/caret-users
>>
>
>
> _______________________________________________
> caret-users mailing list
> caret-users at brainvis.wustl.edu
> http://brainvis.wustl.edu/mailman/listinfo/caret-users
_________________________________________________________________
Want to do more with Windows Live? Learn “10 hidden secrets” from Jamie.
http://windowslive.com/connect/post/jamiethomson.spaces.live.com-Blog-cns!550F681DAD532637!5295.entry?ocid=TXT_TAGLM_WL_domore_092008
More information about the caret-users
mailing list