This has been extensively discussed on the list previously. I think the
conclusion was that the SNR boost of averaging outweighs the better
characterization of the variance.
Peace,
Matt.
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Arman Eshaghi
Sent: Thursday, December 16, 2010 2:42 PM
To: [log in to unmask]
Subject: Re: [FSL] Averaging DTI acquisitions for TBSS
Thanks Matt for your kind replies. That was a stupid mistake of mine not to
see the pairwise bvectors.
BTW why not feed concatenated big4d file into FDT with appropriate bvec? Is
it ok if I move on with this approach if I'm more comfortable with this?
Thanks for your time
Cheers,
Arman
On Dec 16, 2010, at 10:44 PM, Matt Glasser wrote:
> Just change the settings in fslview to see the volume. If the bvectors
are
> not all the same then you might as well not average. How different are
they
> really?
>
> Peace,
>
> Matt.
>
> -----Original Message-----
> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
> Of Arman Eshaghi
> Sent: Thursday, December 16, 2010 10:58 AM
> To: [log in to unmask]
> Subject: Re: [FSL] Averaging DTI acquisitions for TBSS
>
> Thanks Matt for your reply. Two points here:
>
> 1) fslstats shows 0 4055 which obviously is not blank but not visible
> with fslview with the default contrast.
> 2) This is totally a different story: I looked at original bvec. I
expected
> to have a pair of 31 row vectors (1*3 matrix), but this is not the case.
> There are 62 different row vectors. The DTI protocol on Siemens 3T was
with
> one b0 and 30 directions and 2 averages. Does averaging make sense in this
> case at all?
>
> Cheers,
> Arman
>
> On Dec 16, 2010, at 7:37 PM, Matt Glasser wrote:
>
>> Do fslstats data_averaged -R to see if the file is actually blank.
>>
>> Peace,
>>
>> Matt.
>>
>> -----Original Message-----
>> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
Behalf
>> Of Arman Eshaghi
>> Sent: Thursday, December 16, 2010 4:02 AM
>> To: [log in to unmask]
>> Subject: [FSL] Averaging DTI acquisitions for TBSS
>>
>> Dear FSL experts,
>>
>> Following Matt and Steve's comments for averaging I have done the
> following
>> to my data.
>> My data consists of 30 directions acquired in a single protocol with 2
>> averages, 1 b0 and 30 diffusions for 0-31. The sequence is repeated from
> 31
>> to 62.
>> Given the original eddy current corrected file as big4d_corr and the
> command
>> fslroi <input> <output> <tmin> <tsize>:
>>
>> fslroi big4d_corr av1_corr 0 31
>> fslroi big4d_corr av2_corr 31 31
>> fslmaths av1_corr -add av2_corr -div 2 data_averaged
>>
>> I have two problems here:
>> 1) The final result is blank volumes. Is that because I have not aligned
> the
>> volumes? If yes how should I do this?
>> 2) What should I do with bvec for big4d to be applied to my new averaged
>> data?
>>
>> Any input on this is gratefully appreciated.
>>
>>
>> All the best,
>> Arman
|