Hi
yes, this is what I' expect - data sets with increased amount of head
motion will typically have an increased model order. Estimating more
components in those data sets is useful as these additional
components soak up variability due to uninteresting but existing head
motion effects. As such fixing the number of components to 30 does
sound sub-optimal.
cheers
christian
On 12 Jul 2007, at 14:31, Peter Fried wrote:
> I've noticed that letting MELODIC estimate the number of components
> somewhat decreases the amount of noise found in resting-state
> networks.
> It also seems to break some RSNs into multiple components. We had been
> selecting 30 components but with the automatic dim estimate the
> numbers
> are in the 40-50 range (depending ususally on how much motion there
> is).
> Thanks.
>
> -Pete
>
>> Hi,
>>
>> no, it does not matter if some inputs have different number of
>> components. I always think that the automatic dim estimate is what
>> should be used - are numbers very much hgher than 30?
>> cheers
>> c
>>
>>
>> On 11 Jul 2007, at 21:05, Peter Fried wrote:
>>
>>> Hi All,
>>>
>>> If I am merging the Melodic outputs for multiple subjects in
>>> order to
>>> rerun Melodic, do all the inputs have to have the same number of
>>> components?
>>>
>>> I am trying to decide between choosing 30 components for each
>>> subject (180
>>> volumes, TR=2.0) or letting Melodic estimate the number.
>>>
>>> Thanks.
>>>
>>> -Pete
>>
>> ____
>> Christian F. Beckmann
>> University Research Lecturer
>> Oxford University Centre for Functional MRI of the Brain (FMRIB)
>> John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK.
>> [log in to unmask] http://www.fmrib.ox.ac.uk/~beckmann
>> tel: +44 1865 222551 fax: +44 1865 222717
>>
____
Christian F. Beckmann
University Research Lecturer
Oxford University Centre for Functional MRI of the Brain (FMRIB)
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK.
[log in to unmask] http://www.fmrib.ox.ac.uk/~beckmann
tel: +44 1865 222551 fax: +44 1865 222717
|