Dear Paymon,
I'm glad to hear that things are consistent between the different
levels of
Feat analysis. As you only talked about the second and third levels
originally, I was concerned that the first level was different.
So given that the first level shows incorrect orientation, then this
indicates that your original image reconstruction is in error. The
commands that you have applied (avwswapdim *and* avworient)
will change the order of storage in the file, but will *not* change
any actual errors in orientation. Hence I do expect the images
before and after to look identical in FSLView.
What you need to do is to change either the data storage order
(via avwswapdim) *or* the nifti matrix orientation information
(via avworient) but not both. In order to make your images have
"radiological order" - which is best for FSL - you should do the
following:
avworient -getorient IMAGE
If this says Radiological then do:
avwswapdim IMAGE -x y z IMAGE
If it says Neurological then do:
avworient -swaporient IMAGE
Then you should have an image which is stored "radiologically"
(check this with avworient -getorient) and where FSLView
correctly displays left and right.
Once you've done this and checked that it is OK, you will
need to re-run all your analyses again (note that you can
load the previous design matrices in - design.mat in the .feat
directories - and then just modify the input image to point
to the corrected image).
If you have problems with any of this please let us know.
All the best,
Mark
On 16 Oct 2006, at 08:22, Paymon Hosseini wrote:
> Dear Mark,
>
> i ran both "avwswapdim input -x y z output" and "avworient -
> swaporient output" on one fMRI volume. but, when i look at the
> flipped one, and the original one in fslview, there is no
> difference (picture at http://web.mit.edu/paymon/www/
> neurological_vs_radiological_fsl.bmp). this does not seem normal.
> any idea on how i should fix this?
> thanks a lot,
> paymon
>
>
>
>
> Paymon Hosseini wrote:
>> Dear Mark,
>> The fMRI results are consistent across level1, level2, and level3:
>> activations are always on the right side when they should be on
>> the left side. Hence, i doubt there is a problem in the processing
>> stream.
>>
>> Just to double check, what i need to do is run
>> 1) avwswapdim input -x y z output
>> 2) avworient -swaporient output
>> on all initial volumes (structural and functional) and re-run the
>> analysis?
>>
>> thanks a lot,
>> paymon
>>
>> ps.
>>
>> To be complete, here is the output of avwhd for the same volume on
>> which i ran mri_info.
>> >> avwhd enc_1_fmri.nii.gz
>> filename enc_1_fmri.nii.gz
>>
>> sizeof_hdr 348
>> data_type INT16
>> dim0 4
>> dim1 64
>> dim2 64
>> dim3 25
>> dim4 208
>> dim5 1
>> dim6 1
>> dim7 1
>> vox_units mm
>> time_units s
>> datatype 4
>> nbyper 2
>> bitpix 16
>> pixdim0 0.0000000000
>> pixdim1 3.1250000000
>> pixdim2 3.1250000000
>> pixdim3 5.0000000000
>> pixdim4 2.0000000000
>> pixdim5 1.0000000000
>> pixdim6 1.0000000000
>> pixdim7 1.0000000000
>> vox_offset 352
>> cal_max 0.0000
>> cal_min 0.0000
>> scl_slope 0.000000
>> scl_inter 0.000000
>> phase_dim 0
>> freq_dim 0
>> slice_dim 0
>> slice_name Unknown
>> slice_code 0
>> slice_start 0
>> slice_end 0
>> slice_duration 0.000000
>> time_offset 0.000000
>> intent Unknown
>> intent_code 0
>> intent_name
>> intent_p1 0.000000
>> intent_p2 0.000000
>> intent_p3 0.000000
>> qform_name Scanner Anat
>> qform_code 1
>> qto_xyz:1 -3.124882 0.002344 -0.043309 98.407486
>> qto_xyz:2 0.002344 -3.078467 -0.859630 150.438263
>> qto_xyz:3 -0.027068 -0.537269 4.925359 -16.791046
>> qto_xyz:4 0.000000 0.000000 0.000000 1.000000
>> qform_xorient Right-to-Left
>> qform_yorient Anterior-to-Posterior
>> qform_zorient Inferior-to-Superior
>> sform_name Scanner Anat
>> sform_code 1
>> sto_xyz:1 -3.124881 0.000000 -0.043635 98.407486
>> sto_xyz:2 0.004689 -3.078467 -0.859615 150.438263
>> sto_xyz:3 -0.026864 -0.537278 4.925360 -16.791046
>> sto_xyz:4 0.000000 0.000000 0.000000 1.000000
>> sform_xorient Right-to-Left
>> sform_yorient Anterior-to-Posterior
>> sform_zorient Inferior-to-Superior
>> file_type NIFTI-1+
>> file_code 1
>> descrip FreeSurfer Aug 13 2006
>> aux_file
>>
>>
>>
>>
>>
>> Mark Jenkinson wrote:
>>> Dear Paymon,
>>>
>>> Is it the case that the first level results are correctly labeled?
>>> There really should not be a change between the first and second
>>> level results (or third level for that matter). If this is
>>> happening
>>> then something seriously wrong is occurring.
>>>
>>> I'm also not sure why "it seems like the data are in radiological
>>> convention".
>>> FSLView always displays images in this convention (as much as the
>>> original orientation will allow) but what we mean by "radiological"
>>> and "neurological" is about data ordering (see FSL FAQ) and the
>>> result returned by avworient is the true answer to this.
>>>
>>> Unfortunately, I am not familiar with the output from mri_info
>>> so I cannot comment on these. Output from avwhd is more
>>> useful to us.
>>>
>>> The bottom line though is that if your images are not oriented
>>> correctly, then you need to fix this. If it occurs in all levels of
>>> Feat then you probably just need to fix the original images
>>> (and re-run the analyses). If it is OK at first level but then
>>> goes wrong then something is not working correctly and we
>>> will need to find the problem. If this is the case, please tar
>>> up your feat directory (and input files) and send them to
>>> us via our upload site at:
>>> http://www.fmrib.ox.ac.uk/cgi-bin/upload.cgi
>>> and send us the reference number of the upload.
>>>
>>> All the best,
>>> Mark
>>>
>>>
>>>
>>>
>>>
>>>
>>> On 16 Oct 2006, at 02:42, Paymon Hosseini wrote:
>>>
>>>> Dear Mark,
>>>>
>>>> I have decided that my fMRI volumes are left/right flipped
>>>> because my stimuli are verbal, yet all the 3rd level activations
>>>> in FSL are on the right side of feat report images (tagged with
>>>> "R", hence on the right side of the brain). It seems like the
>>>> data are in radiological convention, but they are incorrectly
>>>> tagged as being in neurological convention. Consistent with this
>>>> idea, running "avworient -getorient" results in a "NEUROLOGICAL"
>>>> output.
>>>>
>>>> what i still don't understand is if "it is absolutely necessary"
>>>> to go back to the original volumes, flip them, and redo the
>>>> entire analysis.
>>>>
>>>> thank you very much for your help,
>>>> paymon
>>>>
>>>> ps.
>>>>
>>>> here is the information about about one fMRI slice before i fed
>>>> it (and its comrades) to mri_convert to be converted to nii format:
>>>> -------------
>>>> >> mri_info f_019.bshort
>>>> Volume information for f_019.bshort
>>>> type: bshort
>>>> dimensions: 64 x 64 x 25 x 208
>>>> voxel sizes: 3.1250, 3.1250, 5.0000
>>>> type: SHORT (4)
>>>> fov: 200.000
>>>> xstart: -100.0, xend: 100.0
>>>> ystart: -100.0, yend: 100.0
>>>> zstart: -62.5, zend: 62.5
>>>> TR: 2000.00 msec, TE: 30.00 msec, TI: 30.00 msec,
>>>> flip angle: 90.00 degrees
>>>> nframes: 208
>>>> ras xform present
>>>> xform info: x_r = -1.0000, y_r = 0.0000, z_r = -0.0087,
>>>> c_r = -2.1341
>>>> : x_a = 0.0015, y_a = -0.9851, z_a = -0.1719,
>>>> c_a = 41.3322
>>>> : x_s = -0.0086, y_s = -0.1719, z_s = 0.9851,
>>>> c_s = 26.7234
>>>> Orientation : LPS
>>>> Primary Slice Direction: axial
>>>>
>>>> voxel to ras transform:
>>>> -3.1249 0.0000 -0.0436 98.4075
>>>> 0.0047 -3.0785 -0.8596 150.4383
>>>> -0.0269 -0.5373 4.9254 -16.7910
>>>> 0.0000 0.0000 0.0000 1.0000
>>>>
>>>> voxel-to-ras determinant 48.8281
>>>>
>>>> ras to voxel transform:
>>>> -0.3200 0.0005 -0.0028 31.3708
>>>> 0.0000 -0.3152 -0.0550 46.4996
>>>> -0.0017 -0.0344 0.1970 8.6526
>>>> 0.0000 0.0000 0.0000 1.0000
>>>> -------------------
>>>>
>>>>
>>>>
>>>> here is the information about the nii volume that includes the
>>>> above slice
>>>>
>>>> -------------------
>>>> >>avworient -getorient enc_1_fmri.nii.gz
>>>> NEUROLOGICAL
>>>>
>>>> >> mri_info enc_1_fmri.nii.gz
>>>> INFO: using NIfTI-1 sform
>>>> Volume information for enc_1_fmri.nii.gz
>>>> type: nii
>>>> dimensions: 64 x 64 x 25 x 208
>>>> voxel sizes: 3.1250, 3.1250, 5.0000
>>>> type: SHORT (4)
>>>> fov: 64.000
>>>> xstart: -32.0, xend: 32.0
>>>> ystart: -32.0, yend: 32.0
>>>> zstart: -12.5, zend: 12.0
>>>> TR: 2000.00 msec, TE: 0.00 msec, TI: 0.00 msec, flip
>>>> angle: 0.00 degrees
>>>> nframes: 208
>>>> ras xform present
>>>> xform info: x_r = -1.0000, y_r = 0.0000, z_r = -0.0087,
>>>> c_r = -2.1341
>>>> : x_a = 0.0015, y_a = -0.9851, z_a = -0.1719,
>>>> c_a = 41.3322
>>>> : x_s = -0.0054, y_s = -0.1719, z_s = 0.9851,
>>>> c_s = 26.7234
>>>> Orientation : LPS
>>>> Primary Slice Direction: axial
>>>>
>>>> voxel to ras transform:
>>>> -3.1249 0.0000 -0.0436 98.4075
>>>> 0.0047 -3.0785 -0.8596 150.4383
>>>> -0.0168 -0.5373 4.9254 -17.1134
>>>> 0.0000 0.0000 0.0000 1.0000
>>>>
>>>> voxel-to-ras determinant 48.8268
>>>>
>>>> ras to voxel transform:
>>>> -0.3200 0.0005 -0.0028 31.3708
>>>> -0.0002 -0.3152 -0.0550 46.4992
>>>> -0.0011 -0.0344 0.1970 8.6538
>>>> 0.0000 0.0000 0.0000 1.0000
>>>>
>>>> ------------------
>>>>
>>>>
>>>>
>>>> Mark Jenkinson wrote:
>>>>> Dear Paymon,
>>>>>
>>>>> How are you deciding that the images have been flipped?
>>>>> This is non-trivial to determine by looking at the images, and
>>>>> there has been quite a lot of confusion in the past about
>>>>> how to interpret image orientation. We now provide
>>>>> labels in FSLView when the appropriate qform/sform settings
>>>>> are contained in the nifti file. These labels are the best way
>>>>> to tell how FSL is interpretting your images. We are very
>>>>> careful internally to make sure that left-right is not flipped
>>>>> during the processing, but we do register to the standard
>>>>> space (or highres) images which can involve a large rotation
>>>>> which can appear like a flip at first glance.
>>>>>
>>>>> Also, there are some problems with having images in
>>>>> "neurological order" (see the FSL FAQ for the definition)
>>>>> and if this is the case for your input images then I would
>>>>> suggest changing them to "radiological order" as described
>>>>> in the FSL FAQ.
>>>>>
>>>>> If you still think that there is an incorrect flip then please
>>>>> provide us some details about the images and how you
>>>>> have decided that a flip exists.
>>>>>
>>>>> All the best,
>>>>> Mark
>>>>>
>>>>>
>>>>>
>>>>> On 14 Oct 2006, at 23:28, Paymon Hosseini wrote:
>>>>>
>>>>>> Dear FSL users,
>>>>>>
>>>>>> i apologize beforehand, because you have probably seen a
>>>>>> variation of this question a million times.
>>>>>>
>>>>>> after getting to the second level of Feat multisession/
>>>>>> multilevel analysis, i realized that the left/right
>>>>>> orientations must have been flipped in my original volumes
>>>>>> (converted from COR to nii using mri_convert). Because i am
>>>>>> certain the volumes are flipped, I am wondering if it is
>>>>>> necessary to reorient the data an repeat the analysis from
>>>>>> scratch? Specifically, does this flipped orientation influence
>>>>>> the registration of the original volumes to the standard avg
>>>>>> brain, or is the orientation just a label?
>>>>>>
>>>>>> thanks a lot,
>>>>>> Paymon
|