Mark,
I have been following the orientation discussion with great interest. I am
not clear how I should set qform and sform when I convert dicom files into
nifti. Do you have examples on oblique dicom slices in axial, sagital, and
coronal plane converted to nifti? If not - would you be willing to convert
three data sets for me from dicom to nifti.
I am not clear on a) whether I should set both qform and sform from dicom
headers, and b) whether I should reorder the data if the original slices
were say in sagital format to resemble axial slices and then set the
appropriate qform and sform.
Thanks,
Arvind
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Mark Jenkinson
Sent: Saturday, April 29, 2006 3:40 AM
To: [log in to unmask]
Subject: Re: [FSL] Yet another orientation question
Dear Jolinda,
We do have plans to offer viewing options based on using the
sform or qform to reorient but this won't be available until the
next major release (several months at least). I do think that
we are consistent with almost all other Analyze-style viewers for
the native slicings though, so being able to reconstruct dicom
to nifti in a way that is more consistent with how the Analyze
reconstructions used to be for the native data is, I think, a
valuable feature and is probably useful for other packages in
addition to FSL.
It is true that not all FSL programs support unsigned 16 bit integers
at present, although many do. I think that your work around at
the moment is sensible.
I can't think of any other things that you need to know.
All the best,
Mark
On 28 Apr 2006, at 17:29, jolinda wrote:
> Hi,
>
> I have one last, relatively minor, question. Make that two questions.
> FEAT
> uses the native slicings. However, if a file is converted directly from
> DICOM to NIfTI, the native slicings might be expected to correspond to
> the
> raw data order in the DICOM file, which (ignoring R/L for now) is
> inverted
> top-to-bottom compared to the Analyze order. Another way to state this
> is
> that most DICOM viewers plot the data starting at the upper-left
> corner of
> the view, and most Analyze viewers start plotting from the bottom. So,
> in
> the FEAT report, axial slices will be nose-down, sagittal slices will
> be
> head-down, etc. I was just wondering if there were any plans to
> include a
> flag to switch this, or to use the qform/sform data to detect which
> way to
> plot the data.
>
> I've modified my conversion program to produce an FSL-specific NIfTI
> output,
> as well as a regular NIfTI output. The FSL-specific version differs in
> the
> following ways:
>
> 1. Unsigned 16-bit data is flagged as the signed-16-bit data type,
> rather
> than the unsupported unsigned-16-bit data type. This is acceptable for
> most
> (all that I've seen, anyway) MR data, as it uses only the first 12
> bits.
>
> 2. The data is flipped along the row axis (LR for axial data) and
> rotated
> 180 degrees in the image plane. This is equivalent to a top-bottom
> flip, if
> you prefer to think of it that way. The net result is image data
> stored in a
> left-handed system, LAS for axial images. This will eliminate any need
> for
> FSL to invert the data later, and will give a "normal" view in the FEAT
> report. The qform fields should indicate the relationship between the
> data
> ordering and the NIfTI RAS space.
>
> Is there anything else I should be aware of regarding FSL's use of the
> NIfTI
> format?
>
> Jolinda
>
> -----Original Message-----
> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
> Behalf
> Of Mark Jenkinson
> Sent: Saturday, April 22, 2006 2:02 AM
> To: [log in to unmask]
> Subject: Re: [FSL] Yet another orientation question
>
>
> Hi,
>
> As Steve said, we try to display in the native slicings in FEAT. We
> will,
> however, swap the x-axis in both FSLView and slices (used by FEAT to
> generate the report) if the data is "neurologically" stored. So this
> means
> that you'll see a "radiological" view of the data,
> although
> the current version should helpfully put right-hand markers (small
> "R"s) in
> the images to make it clear where the right-hand side is. This will be
> correctly interpreted from the qform or sform information regardless of
> whether it is "radiologically" or "neurologically" stored.
|