Hi Michael,
Your data is in what *we* have termed "neurological order".
What this means is that the data is stored with a coordinate
system that is a reflected version of the coordinate system
used for the avg152 images (which we term "radiological order").
The issue of reflection is important because our registrations
will never include a reflection (as we don't want to get left-right
mixed up) and so this must be correct before doing registration,
but because images can be rotated by any amount, including
180 degrees, prior to registration, it becomes difficult to easily
and intuitively see which images are "radiologically ordered"
and which are "neurologically ordered". If they are *rotated*
to have the same A-P and S-I axis conventions as the avg152
images then it boils down to the easy case where "radiological
order" stores the right-hand voxels first and left-hand last
while "neurological order" is the other way around. Note
that this is about how the data is stored on disk, and *not* how
the images appear in FSLView. As FSLView in this case
(matching A-P and S-I with the avg152) will always show them
in the radiological convention with right-hand brain voxels on
the left-hand side of the screen.
We still have some minor bugs to sort out with "neurologically
ordered" images, so for this reason I recommend using
the "radiologically ordered" version which I showed you how
to create in the last email. This is an easy process to do
to any images prior to using FSL and so shouldn't really cause
much hassle. I believe that the bugs are limited to reporting
numerical coordinates and processing coordinate values as
inputs, but I'm not 100% sure about this, so I really, really do
not recommend using "neurologically ordered" data.
Hope this helps to explain this a bit more without causing
any brain twisting for you or your images! :)
All the best,
Mark
On 12 Jul 2006, at 20:22, Michael Scheel wrote:
> Hi Mark,
>
> thanks a lot for your help - the way you describe works.
>
> Some questions i still do have to understand what i am actually doing.
> Do i understand you right that fsl interprets my data as being
> neurological orientated and if so why?
> Do all fsl-features work as long as the explicit labels in fslview
> of my data (left and right etc.) match what I expect them to be?
> What would happen if i would use just the converted volumes without
> using the avw-tools?
>
> Sorry for bothering you with this kind of stuff, but this
> orientation issue really twists my brain - Michael.
>> The difficulties that you are having with this image may be
>> because (i) you have the old fslview which does not label
>> the axes - in which case you should upgrade, and/or (ii)
>> that you are expecting fslview to display the left-hand side
>> of the image on the right-hand side of the screen. The
>> latter is a source of confusion, as if the anterior-posterior
>> orientation is not top-bottom (and yours is bottom-top)
>> then left-right are shown on the real left-right of the screen
>> and not reversed. This is so that everything is rotationally
>> consistent. As this is a great source of confusion (and we
>> only refer to our own definition of "radiologically ordered"
>> which does not technically relate to left-right positions on the
>> screen) we have included the explicit labels in fslview so that
>> you can check whether fslview has interpreted the axes
>> correctly. It is really only these labels that you need to worry
>> about, and in this case it seems fine.
>>
>> If you'd rather see the image in the same orientation as
>> the avg152 images, then you can do this by swapping axes.
>> Also, as we do not presently recommend the use of
>> "neurologically oriented" data, as you have here, with
>> FSL, then I suggest you do the following, which will also give
>> you images in the same general orientation as the avg152 images
>> which will be easier to interpret the view:
>> avwswapdim AHconv x -y z AHconvFSL
>> avworient -swaporient AHconvFSL
>>
>> It will warn you about flipping the orientation, but this is what
>> you want, and when you load the image AHconvFSL into FSLView
>> you will see that they have Anterior-Posterior as top-bottom and
>> Left-Right on the opposite sides of the screen. These are,
>> again, all labeled by FSLView and if the labels are incorrect
>> then something has gone wrong.
>>
>> I hope this helps.
>>
>> All the best,
>> Mark
>>
>>
>>
>>
>> On 4 Jul 2006, at 18:28, Michael Scheel wrote:
>>
>>> Hi doug,
>>>
>>> tkmedit displays the volume in the "correct" way. Thanks for your
>>> reply, anyway.
>>>
>>> Michael.
>>>
>>>
>>> Doug Greve schrieb:
>>>>
>>>> Hi Michael,
>>>>
>>>> to make sure that it is not an mri_convert problem, can you load
>>>> the
>>>> volume into tkmedit, ie,
>>>>
>>>> tkmedit -f f.nii
>>>>
>>>> When looking at an axial slice, the nose should be pointing up.
>>>> When
>>>> looking at a sagital slice, the nose should be pointing to the
>>>> right. If you have a fiducial or can tell left from right in some
>>>> other way, then make sure that the left side of the subject is
>>>> on the
>>>> right side of the image. If any of these things are wrong, then
>>>> let me
>>>> know and I will fix mri_convert right away. Otherwise, I'll have to
>>>> pass the buck to FSL ...
>>>>
>>>> cheers
>>>>
>>>> doug
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> On Mon, 3 Jul 2006, Michael Scheel wrote:
>>>>
>>>>> Hi fsl experts,
>>>>>
>>>>> i used mri_convert (freesurfer) to convert my Dicom files
>>>>> (T1-3D-Fiesta-Sequence GE scanner) into nifti format to use fsl
>>>>> on them. When looking at the results with fslview the former
>>>>> radiological oriented volumes seem to be rotated 180° around
>>>>> the z-axis i.e. nose pointing down, right-ear on right side of
>>>>> the screen, left on the left.
>>>>> Running avworient -getorient does return NEUROLOGICAL.
>>>>> I am trying to solve this problem for days now including using
>>>>> different converters and reading carefully the faqs about
>>>>> neurological, radiological, qform and sform. But I have to
>>>>> admit that I rather get more confused than having the feeling
>>>>> to get somewhat nearer to solve this. May just somebody take a
>>>>> look at an example of my data and tell me what to do?
>>>>>
>>>>> Thanks, Michael.
>>>>>
>>>>>
>>>>
>>>
>>> --Michael Scheel
>>> AG PD Dr. Stroehle
>>> Klinik fuer Psychiatrie und Psychotheraphie
>>> Charite - Campus Charite Mitte
>>> [log in to unmask]
>>> 0176 - 20801852
>>
>
> --
> Michael Scheel
> AG PD Dr. Stroehle
> Klinik fuer Psychiatrie und Psychotheraphie
> Charite - Campus Charite Mitte
> [log in to unmask]
> 0176 - 20801852
|