Hi,
See my previous email for the first point.
I would not smooth the data at all.
It is theoretically nicer to unwarp prior to averaging,
so I'd recommend that although others may have
something to add, as I haven't done this myself before.
All the best,
Mark
On 12 May 2009, at 10:25, Siew-Min Gan wrote:
>> Thanks. May I ask you a few more questions:
>
> I got the undistorted 4D DTI brain output from FEAT GUI
> (filtered_func_data.nii.gz) which looks good. However, it is
> deskulled and
> has less brain tissue compared to the undistorted 3D DWI brain output
> which is not deskulled (EF_UD_example_func.nii.gz in unwarp
> subfolder).
> The loss of brain tissue due to masking is quite visible in the
> inferior
> temporal lobe, which is the region I'm interested in. I have betted
> the
> magnitude file very liberally but the problem remains. I assume this
> is
> used to mask/bet the 4D data. Is there a default 4D undistorted brain
> output which is not deskulled from running the B0 unwarping function
> in
> FEAT GUI, as I couldn't locate it? If there isn't, what other ways
> can be
> done to lessen the tissue lost in the undistorted 4D DWI brain output?
>
> Secondly, is there a benefit to apply FWHM spatial smoothing when
> carrying
> out only the B0 unwarping function on DTI data?
>
> Lastly, if I have two repetition/ acquisitions of 4D DWI, do you
> suggest
> it's better to perform the B0 undistortion on them seperately, before
> averaging them?
>
> Thanks
>
> Siewmin
>
> Hi,
>>
>> I'm not sure why the intensity range is different in your nifti
>> image, but use that range. It is more likely that the reconstruction
>> software tries to use the range 0 to (2^n - 1) for the intensity
>> range. Hence you should map the range 2048 to 2*pi. You
>> could also argue that 2047 is better - but the difference, which
>> is less than 0.05% won't affect anything significantly, given
>> the typical SNR in MR images.
>>
>> All the best,
>> Mark
>>
>>
>>
>>
>> On 11 May 2009, at 21:39, Siew-Min Gan wrote:
>>
>>> Hi, on the fsl practical page, there is a step on preparing phase
>>> difference map to a fieldmap in rad/sec with the following command
>>> "fslmaths phasemap -sub 2048 -div 2048 -mul 3.14159 phase_rad -odt
>>> float".
>>>
>>> When I look at the histogram of the intensity of the phasedif map,
>>> the
>>> dicom file has a range of intensity from -4096 to 4092. In nifti
>>> format,
>>> it's intensity range is 0 to 4094. May I ask if it is more accurate
>>> to
>>> keep to 2048 or to substitute 2048 to 2047 in the above calculation?
>>>
>>> Thanks
>>>
>>> Siewmin
>>>
>>
>
|