Hi Andreas,
yes, I think it would probably be wise to be very careful with using the current release. If for no other reason that it will be hard to convince reviewers given that there is no paper validating it published yet. I thought I had hidden the help text for the —repol option, but clearly I must have missed that.
And yes you are correct, the *.eddy_outlier_free_data.nii.gz file is the one where *only* outlier replacement has been done whereas the regular output is the one where everything has been done.
Jesper
On 26 Feb 2015, at 14:21, Andreas Bartsch <[log in to unmask]> wrote:
> Dear Jesper,
>
> would you then advice to refrain from using --repol with the current
> release, and - if one uses it - which is the eddy-corrected output >with<
> outliers removed?
> (Ruskin seems to indicate that data_eddy.nii.gz appears the eddy-corrected
> output without outlier removal while
> data_eddy.eddy_outlier_free_data.nii.gz does not appear to be corrected
> for eddy currents.)
> Hope all is well,
> cheers-
> Andreas
>
> Am 26.02.15 12:03 schrieb "Jesper Andersson" unter
> <[log in to unmask]>:
>
>> Dear Ruskin,
>>
>> that misreporting is due to a bug in the reporting itself where the
>> numbers reported are between the predictions and the predictions of the
>> previous iteration, rather than between the predictions and the
>> observations. It still uses the correct numbers (which should be above 4
>> with the default settings) for the decisions.
>>
>> Having said that, I would be a little careful with using the outlier
>> functionality just yet. A lot of work has gone into that since the last
>> release and we are now attempting to properly validate it.
>>
>> Jesper
>>
>> On 25 Feb 2015, at 15:29, R Hunt <[log in to unmask]> wrote:
>>
>>> I see (after trying it) that --repol produces the following output for
>>> --out=data_eddy:
>>> 1. data_eddy.nii.gz
>>> 2. data_eddy.eddy_parameters
>>> 3. data_eddy.eddy_outlier_free_data.nii.gz
>>> 4. data_eddy.eddy_outlier_report
>>>
>>> However, I'm confused by the output.
>>>
>>> File 1 appears to be eddy corrected output, but I assume that it has
>>> not had any outliers removed. Am I correct?
>>>
>>> File 3 does NOT appear to have been eddy corrected and seems identical
>>> to the --imain input data. Am I correct about this as well?
>>>
>>> File 4 lists all outliers as being less than +/-0.25 "standard
>>> deviations off". Are these slices actually outliers? If so, how is this
>>> reflected in the SD value?
>>> Slice 3 in scan 50 is an outlier -0.0149151 standard deviations off
>>> Slice 8 in scan 38 is an outlier -0.0440188 standard deviations off
>>> Slice 8 in scan 91 is an outlier -0.0904612 standard deviations off
>>> Slice 13 in scan 52 is an outlier 0.245672 standard deviations off
>>> Slice 36 in scan 114 is an outlier -0.00400951 standard deviations
>>> off
>>>
>>> Can someone explain what's going on?
>>>
>>> The full command was:
>>> eddy --imain=data --mask=mask --acqp=topup_datain.txt
>>> --index=eddy_index.txt --bvecs=bvecs.bvec --bvals=bvals.bval
>>> --flm=quadratic --topup=topup_output --out=data_eddy --verbose
>>>
>>> Thanks,
>>> Ruskin
|