Hi Mark,
Thanks for getting back. Just out of curiosity. Pursuing this is
entirely optional.
$ fslstats SE3.nii -R
-850.000000 1160.000000
$ fslhd SE3.nii |grep cal
cal_max 0.0000
cal_min 0.0000
$ fslhd SE3.nii |grep scl
scl_slope 2.000000
scl_inter -4096.000000
Tools > Image Histogram > Use log scale. On the histogram the lowest and
the highest clickable bins are -858 and 1157, respectively.
Default Min=-4670.4 Max=-2990, I see three whole-white views.
When set Min=-900 Max=-851, I see three whole-white views.
When set Min=-850 Max=1160, I see a brain.
When set Min=1161 Max=1200, I see three whole-black views.
In all cases (X,Y,Z) = (dim1/2,dim2/2,dim3/2) so there is no issue of
cutting the wrong plane where there's no brain.
Cheers,
mary
On 2017-03-10 06:54, Mark Jenkinson wrote:
> Hi,
>
> There are some heuristics built into FSLView to calculate the default
> display range and these can go wrong. However, it could also be due
> to values stored in the nifti header, as if cal_min and cal_max are
> not zero (you can see these values by using fslhd) then FSLView will
> use these values, as they are intended to be a suggested display
> range, although sometimes they are set poorly and that causes
> problems. If this is the problem then you can use fsledithd to get
> rid of them (set them both to zero).
>
> As for scl_slope and scl_inter, all FSL tools, including FSLView,
> should honour these and use them. So nothing, as far as I am aware,
> should be using non-scaled intensity values.
>
> All the best,
> Mark
>
>
>
>> On 4 Mar 2017, at 06:46, Mary Chin <[log in to unmask]> wrote:
>>
>> Hi,
>>
>> Just trying to understand how FSLView decides on the default
>> (brightness) window. Is there anything to do with scl_slope and
>> scl_inter?
>>
>> Asking, because with some nii I get:
>> - three whole-white ortho views;
>> - brightness Min and Max that are completely outside the range of
>> pixel values in the nii.
>>
>> No major issue here, because once I change the 'Min' and 'Max'
>> manually to the min and the max, voila the whole-whites disappear and
>> I get the expected display. The min and the max I input are consistent
>> with:
>> - the output from 'fslstats -R';
>> - the output from Image Histogram.
>> These pixel values are without linear remapping using scl_slope and
>> scl_inter. (Which I understand to be correct, as the scl_slope and
>> scl_inter values in the file are aberrations from dicom-to-nii
>> conversion, which is a separate matter).
>>
>> Wondering how come a different set of min-max is used for the default
>> window.
>>
>> Thanks,
>> mary
|