I see.
Yes we talked about how to choose the number of ICs before. I read the paper from Minka, MIT. But to be honest to you, that paper I dont quite get that:(
Thanks Christian:)
Xuelin
****************************************
Xuelin Cui
Department of Electrical Engineering
University of Hawaii-Manoa
Honolulu HI 96822
Tel: 1-808-349-0983
Email: [log in to unmask]
****************************************
----- Original Message -----
From: Christian Beckmann <[log in to unmask]>
Date: Monday, April 30, 2007 12:54 pm
Subject: Re: [FSL] a question on IC map and time course
To: [log in to unmask]
> Hi,
>
> Yes, unless you change the default settings the number of
> components
> is controlled via an initial PCA as described in the paper and
> technical report (have we not had this discussion before?). The
> threshold parameter is unrelated to all this - it relates to the
> way
> that the spatial IC maps get thresholded when using the Gaussian/
> Gamma mixture model in order to perform an alternative hypothesis
> test: voxels with a larger posterior probability of 'activation'
> are
> shown.
> cheers
> Christian
>
>
>
> On 30 Apr 2007, at 23:22, Xuelin Cui wrote:
>
> > Christian:
> >
> > Thanks. With your reply and the melodic _slices, I have more
> sense
> > about ICA on fMRI now. One more question: Is the number of the
> > independent components controlled by preprocess PCA? There is a
> > parameter that I can modify in MELODIC, threshold. Is that
> > parameter for controlling the summation of eigenvalues in PCA?
> Thus
> > the number of IC afterwards?
> >
> > Thanks a lot
> >
> > Xuelin
> >
> > ****************************************
> > Xuelin Cui
> > Department of Electrical Engineering
> > University of Hawaii-Manoa
> > Honolulu HI 96822
> >
> > Tel: 1-808-349-0983
> > Email: [log in to unmask]
> > ****************************************
> >
> > ----- Original Message -----
> > From: Christian Beckmann <[log in to unmask]>
> > Date: Monday, April 30, 2007 4:00 am
> > Subject: Re: [FSL] a question on IC map and time course
> > To: [log in to unmask]
> >
> >> Hi
> >>
> >> I do not understand why you start talking about slices - the rest
> >> is
> >> bbasically right.
> >> cheers
> >> c
> >>
> >>
> >>
> >>
> >> On 30 Apr 2007, at 14:50, Xuelin Cui wrote:
> >>
> >>> Christian:
> >>>
> >>> Thanks for your reply.
> >>>
> >>> According to your last Email, can I understand the spatial map
> >> and
> >>> the time course for a single component as follows: Those spatial
> >>> maps of a single IC is aquired at same time, but from different
> >>> slices. The spatial map of an single IC is for all the time
> >> points
> >>> during the scan. The corresponding time course is for controlling
> >>
> >>> how the spatial map contribute to the fMRI signal at each time
> >> points.>
> >>> Am I right?
> >>>
> >>> Thanks a lot
> >>>
> >>> Xuelin
> >>>
> >>> ****************************************
> >>> Xuelin Cui
> >>> Department of Electrical Engineering
> >>> University of Hawaii-Manoa
> >>> Honolulu HI 96822
> >>>
> >>> Tel: 1-808-349-0983
> >>> Email: [log in to unmask]
> >>> ****************************************
> >>>
> >>> ----- Original Message -----
> >>> From: Christian Beckmann <[log in to unmask]>
> >>> Date: Monday, April 30, 2007 7:53 pm
> >>> Subject: Re: [FSL] a question on IC map and time course
> >>> To: [log in to unmask]
> >>>
> >>>> Hi,
> >>>>
> >>>> On 30 Apr 2007, at 08:54, Xuelin Cui wrote:
> >>>>
> >>>>> hi folks:
> >>>>>
> >>>>> I ran MELODIC for fMRI data, and thus have a bunch of
> >>>> independent
> >>>>> components which are displayed by a html file. In the html file,
> >>>>
> >>>>> there is a spatial map, a time course, as well as a Fourier
> >>>>> spectrum of the time course.
> >>>>>
> >>>>> I have some questions on those in the html file:
> >>>>>
> >>>>> 1. Are all the spatial maps in single html file snapped at
> >>>>> different time point? IF so what time are they?
> >>>>>
> >>>> No
> >>>>
> >>>>> 2. What is the relation between the time course and the spatial
> >>>> map
> >>>>> in the same html file?
> >>>>>
> >>>>
> >>>> They express the spatial and temporal characteristics of a
> >>>> component
> >>>> - in a similar way that EVs (regressors in the GLM) and the
> >>>> associated maps of parameter estimates express BOLD modulation.
> >>>> That
> >>>> is, the spatial maps corresponds to a map of parameter estimates
> >>>> and
> >>>> the time course describes how that spatial map is modulated in
> >>>> strength at different time points.
> >>>>
> >>>>> 3. Are the 125 points time course variance corresponds to a
> >>>> certain
> >>>>> voxel in the brain? If so, which voxel is it? If not, what is it
> >>>>
> >>>>> related to? Some certain region?
> >>>>>
> >>>>
> >>>> Likewise, the time course is related to the entire IC map -
> not to
> >>>> a
> >>>> single voxel.
> >>>> cheers
> >>>> christian
> >>>>
> >>>>> Thanks a lot
> >>>>>
> >>>>> Xuelin
> >>>>>
> >>>>> ****************************************
> >>>>> Xuelin Cui
> >>>>> Department of Electrical Engineering
> >>>>> University of Hawaii-Manoa
> >>>>> Honolulu HI 96822
> >>>>>
> >>>>> Tel: 1-808-349-0983
> >>>>> Email: [log in to unmask]
> >>>>> ****************************************
> >>>>
> >>>> ____
> >>>> 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
> >>
>
> ____
> 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
>
|