Hi,
If you are unwarping spin-echo EPI images (as for diffusion)
then you do not need to mask out the signal loss areas as
is done for FMRI (gradient-echo EPI). Hence you do not
need to use the TE of the EPI sequence, as that is all it is
used to calculate. However, it does no particular harm to
mask out these areas, so it is not crucial. You can also
achieve the same by setting the "% Signal loss threshold"
to zero in the GUI.
All the best,
Mark
On 3 May 2009, at 00:20, Siew-Min Gan wrote:
> Hi Matt, thanks. I made an error previously and it worked now. With
> the
> command lines you wrote, I noticed there isn't a need for the TE of
> the
> EPI, whereas this parameter is required in the FEAT GUI prestats B0
> unwarping. Why is that so?
>
>
> Thanks
>
> Siewmin
> I would try converting the field maps using dcm2nii, in the MRIcron
>> package,
>> or MRIConvert and see if you get the correct intensities that way
>> (I use
>> dcm2nii and get intensities that work). As for trying to manually
>> scale
>> the
>> images, I am honestly not sure, perhaps someone else on the list
>> knows
>> better.
>>
>> About the IPAT, I think that is correct, but again someone else on
>> the
>> list
>> should verify as that is the kind of question I would tend to ask
>> one of
>> my
>> BME colleagues with more extensive MR physics background. ;)
>>
>> Regarding the "few voxels more" could you send me a screenshot of
>> what you
>> are talking about? Were your brain extractions good? You might
>> need to
>> modify the bet lines in the code I sent to be optimal for your data
>> (specifically the -f option, and -g might need to be set as well).
>>
>> Peace,
>>
>> Matt.
>>
>> -----Original Message-----
>> From: Siewmin Gan [mailto:[log in to unmask]]
>> Sent: Friday, May 01, 2009 12:17 PM
>> To: [log in to unmask]; Matt Glasser
>> Cc: Siewmin Gan
>> Subject: Re: problem running FEAT
>>
>> Hi Matt, thank you very much for your help.
>>
>> Can I ask you a basic question?
>> I assume the 2 GRE maps I have are a realfieldmap and a magnitude
>> map from
>> comparing them to the fsl webpage on how they may look like. The
>> "realfieldmap" dicom
>> has intensity -4096 to + 4094. Am I right to then prepare this
>> realfieldmap
>> by calculating
>> it to make it of a range from either -pie to pie or 0 to 2 pie?
>>
>> With the fieldmaps, they are acquired from a 3T Siemens Scanner and
>> converted from
>> dicom to Analyze/Nifti using Debabeler. I also tried another way to
>> convert
>> them, this
>> time using SPM. The fieldmap output from SPM would retain intensity
>> values
>> from -4096
>> to +4092. I tried to multiply them by 3.14 and divide by 4096 and
>> their
>> intensity range
>> becomes-pie to pie or -3.14 to + 3.14. However, when I input this
>> into the
>> Feat Gui, an
>> error msg pops up with "Error:Child Process Exited Abnormally". May
>> I ask
>> what may be
>> the reason for this? I read there is no difference between the two
>> ranges
>> but the former
>> range fieldmap I have doesn't seem to be accepted on the GUI. Would
>> it
>> matter if the
>> intensity are from -pie to pie using the command lines?
>>
>>
>> I tried the command lines you've written. I noticed a few voxels
>> more in
>> the
>> brain edges
>> in the output undistorted images compared to the original 4D data.
>> I'm not
>> sure if it's due
>> to masking. I tried both y and y- directions and there is no
>> difference in
>> the output. If this
>> is so, could this mean that my input data is faulty, so the output
>> from
>> the
>> command lines
>> didn't undistort the data? Also, with the command lines, if I use
>> IPAT 2
>> parallel processing
>> on the Scanner, should I input 1/2 the dwell time when running this
>> function?
>>
>> I tried the FSL FEAT GUI again now, and there are more command lines
>> coming
>> out of the
>> log file. I'm waiting to see if I could get output files this time.
>> For
>> using the GUI to only
>> do B0 distortion, is it necessary to apply spatial smoothing FWHM
>> and /or
>> Temporal
>> Filtering option on the pre-stats section?
>>
>> Many Thanks again.
>>
>> Siewmin
>>
>
|