This would not work for our analysis. We are using randomise to
compare the spatial distributions of white matter hyperintensities
between groups, using binary images (1-lesion, 0-not). In one group,
they also have had intracranial hemorrhage. We would not want to
code these voxels as 0, as we can not know the condition of the
underlying white matter.
So we want to code missing data by voxel in individual subjects. An
alternative to NaN would be to use -1, but similarly I am not sure
how fsl will deal with this.
Chris
On Jun 16, 2006, at 12:55 AM, Steve Smith wrote:
> Hi - as far as I know, inf/NaN isn't generally-supported in Analyze/
> NIFTI files - and FSL doesn't use (deal nicely with) NaN values.
> Possible workarounds depend on the nature of the missing data - for
> example, if you have a corrupt timepoint, you could include an
> extra EV that contains zeros everywhere and a 1 at that
> timepoint.....but I suspect that this doesn't solve you problem.....
>
> Cheers, Steve.
>
>
> On 16 Jun 2006, at 05:48, Christopher Holland wrote:
>
>> So how would either program deal with voxels that with the value
>> NaN, would it just ignore those points, or would that result in an
>> NaN being generated for that voxel in the result image? There is
>> no good way to fill in the missing data with mean or other data.
>>
>> Thanks,
>>
>> Chris
>>
>> On Jun 16, 2006, at 12:41 AM, Steve Smith wrote:
>>
>>> Hi - I'm afraid there isn't a "missing data" feature coded into
>>> FEAT or randomise, sorry! You might want to workaround by
>>> filling in missing data from existing data, depending on the
>>> details.
>>>
>>> Cheers, Steve.
>>>
>>>
>>> On 15 Jun 2006, at 21:59, Chris Holland wrote:
>>>
>>>> For FEAT or Randomise analysis, how should I code 'missing data'
>>>> in voxels
>>>> for subject images in which we do not have data. Is using NaN
>>>> sufficient or
>>>> does FSL have a particular convention, such as -1 for coding
>>>> missing data
>>>> within images? Thanks,
>>>>
>>>> Chris
>>>
>>>
>>> --------------------------------------------------------------------
>>> -------
>>> Stephen M. Smith, Professor of Biomedical Engineering
>>> Associate Director, Oxford University FMRIB Centre
>>>
>>> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
>>> +44 (0) 1865 222726 (fax 222717)
>>> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
>>> --------------------------------------------------------------------
>>> -------
>
>
> ----------------------------------------------------------------------
> -----
> Stephen M. Smith, Professor of Biomedical Engineering
> Associate Director, Oxford University FMRIB Centre
>
> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
> +44 (0) 1865 222726 (fax 222717)
> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
> ----------------------------------------------------------------------
> -----
|