Sorry, a correction: this small dataset of course does not contain outliers... but real data from my subjects that do also do not yield confound outputs!
On Mon, 21 Oct 2013 12:50:38 +0100, Carlos Pato Preto <[log in to unmask]> wrote:
>Happens every time, and using two different cluster servers (one running on Centos 6.4 and the other SUSE 11). Until last week I was using FSL 5.0.0 on one and 5.0.4 on the other, now updated to 5.0.5 and still the same. The error message I posted before doesn't show up on the terminal, I can only see it when I parallelize the process; that's what comes up in the error file.
>
>As an example, I ran fsl_motion_outliers with the --dvars option in a functional segment with 10 volumes and got these values in the -s file:
>95.1659
>94.7451
>96.3985
>96.202
>94.2215
>93.4431
>93.4543
>93.6447
>94.7123
>
>I suppose values above 96.6871 should be considered outliers, but I get nothing.
>
>Regards,
>
>Carlos
>
>
>On Mon, 21 Oct 2013 11:13:45 +0000, Mark Jenkinson <[log in to unmask]> wrote:
>
>>Hi,
>>
>>This command is fine and should work.
>>I don't understand why you see that error message.
>>Does it happen every time?
>>
>>When I check the calculation of the threshold with the 75th percentile + 1.5 * IQR, I get the same value via the -s output and via the verbose output (and via counting the number of values about this threshold). So can you give some more details about how it is different in your case?
>>
>>Also, what version of FSL are you running, and on what kind of machine?
>>
>>All the best,
>> Mark
>>
>>
>>
>>On 21 Oct 2013, at 11:13, Carlos Pato Preto <[log in to unmask]> wrote:
>>
>>> Hello,
>>>
>>> I'm using the following command (which I hope is wrong ;) ):
>>> fsl_motion_outliers -i func.nii.gz -o confound.txt --dvars
>>>
>>> Thanks!!
>>>
>>> Carlos
>>>
>>>
>>> On Mon, 21 Oct 2013 07:00:18 +0000, Mark Jenkinson <[log in to unmask]> wrote:
>>>
>>>> Hi,
>>>>
>>>> The default metric does have a bias towards choosing the middle timepoints, so that is not unusual.
>>>> I am not sure why you are having trouble with dvars. What is the exact command that you are running?
>>>>
>>>> All the best,
>>>> Mark
>>>>
>>>>
>>>>
>>>> On 17 Oct 2013, at 09:49, Carlos Pato Preto <[log in to unmask]> wrote:
>>>>
>>>>> Hi, FSL experts
>>>>>
>>>>> I've been trying to use fsl motion outliers to clean up my functional data but have come across a couple of issues:
>>>>>
>>>>> 1. When I use dvars as a metric (as I originally intended to), I get no counfound matrix output (although checking the -s file I can find values above the 75th percentile + 1.5 times the interquartile range), and I get the following error:
>>>>> ERROR: Could not open image /tmp/fsl_dYN81j_mc/fmri_mcf
>>>>> Is there an easy way to fix this?
>>>>>
>>>>> 2. When using the default metric, however, I get a counfound matrix which always has at least two columns with 1's in the rows corresponding roughly to the two middle volumes (always the same two rows for all subjects). Is this as it should be?
>>>>>
>>>>> Many thanks in advance!
|