Dear Angela,
The advice is "look at your images".
You need to decide for yourself which are the better segmentations
based on the result images and then use the ones that are more
accurate. There is no substitute, or good alternative, to careful
manual inspection.
All the best,
Mark
On 12 Sep 2011, at 15:07, Angela Favaro wrote:
> Dear FSL masters,
> I have used both FSL First and Freesurfer to obtain volumes of subcortical
> areas of GM.
> The differences between the two methods are large and, especially in the
> patient group (anorexia nervosa), the correlation between the two methods
> is lower than expected (spearman rho about 0.4). Also the differences
> between groups gave different results (First says that hippocampus is
> smaller in AN and Freesurfer that left accumbens is larger in AN in
> comparison to controls and no difference in the hippocampus).
> What can I do?
> Which method is more reliable?
>
> any suggestion is appreciated
>
> Angela
>
>
>
>
>> Hi - you could use fsl_glm to regress the confounds out before running
>> ICA?
>> Cheers.
>>
>>
>> On 12 Sep 2011, at 12:30, Klooster, Debby wrote:
>>
>>> Dear all,
>>>
>>> I have resting state data. The .txt file with my confounds contains 6
>>> columns for motion parameters (I had to calculate the confounds with SPM
>>> since that is the program that is used in the clinic where I work), a
>>> linear trend and a quadratic trend.
>>>
>>> Since the input in MELODIC is a nifti file, I would like to use a Nifti
>>> file in which correction for these confounds is already applied.
>>> (According to standard GLM: X=mu*M+gamma*G+noise, in which M are the
>>> predictors and G the confounders. Mu and gamma are estimated and I would
>>> like to have a corrected version of the data: X*=X-gamma*G).
>>>
>>> Do you know how I can obtain this? Or do you know how motion parameters
>>> or other confounds are usually taken into account when applying ICA on
>>> resting state datasets?
>>>
>>> Thank you!
>>> Debby
>>>
>>>
>>>
>>> -----Original Message-----
>>> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
>>> Behalf Of Benjamin Kay
>>> Sent: Monday, September 12, 2011 12:58 PM
>>> To: [log in to unmask]
>>> Subject: Re: [FSL] Where to preprocess resting state data for ICA?
>>>
>>> On Monday, September 12, 2011 05:34:29 you wrote:
>>>> Dear FSL users,
>>>> I would like to perform ICA on resting state data. However, I would
>>> like to
>>>> correct for some confounders (a .txt file exists of all the
>>> confounders I
>>>> would like to use). When I use FEAT for this, the program asks a model
>>>> setup for a standard GLM analysis, but off course I don't have any
>>>> predictors since it is resting state data. Can somebody please tell me
>>>> where I can make FSL correct for my confounders? Thank you in advance.
>>>> Kind regards,
>>>> Debby
>>>
>>> Dear Debby,
>>>
>>> When I think of confounds, I usually think of high-level statistics. In
>>> what
>>> sense do you think FSL should be correcting for them during
>>> preprocessing?
>>> With regards to statistical inference on the results of ICA you may find
>>> what
>>> you're looking for in the Post-Stats section of the Melodic GUI or else
>>> here:
>>> http://www.fmrib.ox.ac.uk/analysis/dualreg/
>>>
>>> If you tell us a bit more about your experiment, i.e. what are your
>>> confounds,
>>> what kind of ICA are you doing (tensor vs temporal concatenation), etc.,
>>>
>>> someone may be able to give you more specific advice.
>>>
>>> Benjamin
>>>
>>>
>>>
>>> De informatie verzonden in dit e-mailbericht is uitsluitend bestemd voor
>>> de geadresseerde. Openbaarmaking, vermenigvuldiging, verspreiding en/of
>>> verstrekking van de informatie aan derden is, behoudens voorafgaande
>>> schriftelijke toestemming van de afzender niet toegestaan. Ook staat de
>>> afzender niet in voor de juiste en volledige overbrenging van de inhoud
>>> van een verzonden e-mail bericht, noch voor een tijdige ontvangst
>>> hiervan. Indien bovenstaand e-mailbericht niet aan u gericht is,
>>> verzoeken wij u vriendelijk doch dringend contact op te nemen met de
>>> afzender ervan en eventuele kopieën te verwijderen en te vernietigen.
>>>
>>> The information contained in this message and or attachments is intended
>>> only for the person or entity to which it is addressed. Any use,
>>> forwarding, distribution, disclosure, printing or copying of this
>>> information without prior consent from the sender is strictly
>>> prohibited. The sender accepts no responsibility for any damage and/or
>>> consequences of improper and incomplete transmission of the information
>>> contained in this message or delay in its receipt. If you received this
>>> in error, please contact the sender and delete the material from any
>>> system and destroy all copies.
>>>
>>> Email scanned by Checkpoint Anti-Virus
>>>
>>
>>
>> ---------------------------------------------------------------------------
>> Stephen M. Smith, Professor of Biomedical Engineering
>> Associate Director, Oxford University FMRIB Centre
>>
>> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
>> +44 (0) 1865 222726 (fax 222717)
>> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
>> ---------------------------------------------------------------------------
>>
>>
>>
>>
>
|