Dear Zhi,
You can create an SPM.mat file without data from the batch interface:
SPM > Stats > fMRI model specification (design only)
For the VOI file (the xY variable), have a look at spm_regions.m: you
will have to inform the model of how your time series were obtained (in
particular, adjustment for confounds and temporal filtering). As Amir
mentioned, DCM expects the time series to be whitened.
Best regards,
Guillaume.
On 03/12/2018 15:48, Zhi Li wrote:
> Dear Amir,
>
> I know how to create the SPM.mat and VOI.mat files by myself now. Thank
> you very much for your kind suggestions. One more question is how could
> I calculate the X0 matrix? Because I cannot use the 'Volume of Interest'
> module in SPM.
>
> Best wishes,
>
> Zhi
>
>
>
>
> On Mon, 3 Dec 2018 at 08:21, Amirhossein Jafarian
> <[log in to unmask]
> <mailto:[log in to unmask]>> wrote:
>
>
> Dear Zhi,
>
> A structure called SPM, saved in a file called SPM.mat, with the fields:
>
> SPM.xY.RT – The repetition time (secs)
> SPM.Sess(1).U(c).dt – Sampling interval of the experimental design
> (typically 1/16 * TR)
> SPM.Sess(1).U(c).u – Vector with one row per time bin
> SPM.Sess(1).U(c).name{1} – Name of the experimental condition
>
> Where c=1…C experimental conditions. You’ll also need one file per
> timeseries containing two things: a vector Y (the timeseries) and
> structure array xY with fields:
>
> xY(i).u – the timeseries (same as Y, above)
> xY(i).name –the name of the region (a string)
> xY(i).X0 – a matrix of uninteresting effects to be regressed out of
> the timeseries. One row per MRI volume and one column per effect.
> xY(i).sess – the run or session of this timeseries (e.g. set this to
> 1 if there’s only one run)
>
> Where i is the region number. Note that DCM assumes that the
> timeseries have been pre-whitened. Armed with the SPM and xY files,
> you can use the GUI to specify the DCM or use spm_dcm_specify.m in
> the latest release of SPM.
>
> You can also look into the spm website and go
> to... spm/data/attention/ which gives a demo for how one can
> implement DCM for fMIR with step by step guide in the manual
> (chapter 35).
>
>
> I hope this would help.
>
> Best regards,
>
> Amir.
>
>
> On Friday, 30 November 2018, 15:04, Zhi Li <[log in to unmask]
> <mailto:[log in to unmask]>> wrote:
>
>
> Dear Amir,
>
> Thank you very much. May I ask that if the X.beta you mentioned is
> the X0 matrix in the xY structure in the VOI.mat extracted with
> SPM12? And the value in each column of X0 represents the mean beta
> value of the stimulus-specified beta.nii file? Because the time
> series we got were from the surface-based fMRI data which should be
> incompatible to the SPM required data format (the first-level
> analysis was executed with FreeSurfer). However, if the X.beta you
> mentioned is the average beta value of this ROI, I could extract it
> from the surface-based fMRI data. I think I could get the time
> series, u matrix and beta series. What's the next step should I
> take? How could I feed the DCM modules in the SPM with these
> informations without editing the codes of the dcm functions? Do you
> have any examples or demo scripts? Looking forward to your kind reply.
>
> Thanks and best wishes,
>
> Zhi
>
> On Fri, 30 Nov 2018 at 04:34, Amirhossein Jafarian
> <[log in to unmask]
> <mailto:[log in to unmask]>> wrote:
>
> Dear Zhi,
>
> If you extract ROIs using SPM12 and also having your u matrices
> you do not need SPM.mat as the X.beta can be calculated using
> information in ROIs. In other words you need 3 sorts of
> information for your DCM from SPM.mat : 1-ROIs, 2-inputs, 3-X.beta.
>
> I hope this would help.
>
> Regards,
> Amir
>
>
> On Thursday, 29 November 2018, 15:40, Zhi Li
> <[log in to unmask] <mailto:[log in to unmask]>> wrote:
>
>
> Dear Amir,
>
> Thank you very much for your kind reply. According to my
> understanding it should work after feeding it with the BOLD
> function /y/, the input matrix /u/ and the connectivity matrix A
> B C D, right? I am not clear if all the parameters in the
> SPM.mat file are necessary to the DCM analysis. Could I create a
> SPM.mat file specific to the DCM that this .mat file will has no
> inputted images and won't been estimated, I mean just a design
> of experiment?
>
> Thanks and best wishes,
>
> Zhi
>
> On Thu, 29 Nov 2018 at 03:07, Amirhossein Jafarian
> <[log in to unmask]
> <mailto:[log in to unmask]>> wrote:
>
> Dear Zhi,
>
> Yes you need to have SPM.mat because it includes inputs that
> one would need for DCM! Additional to that please note that
> the observation BOLD function, y, requires having (linear)
> mixture of confounds (e.g., drift) X(beta) for which you
> need to create the SPM.mat file. You can find a summery of
> these in page 315 of SPM12 manual.
>
> I hope this would help.
>
> Best regards,
>
> Amir
>
>
> ____
>
>
> On Wednesday, 28 November 2018, 19:16, Zhi Li
> <[log in to unmask] <mailto:[log in to unmask]>> wrote:
>
>
> Dear SPM experts,
>
> May I ask how to use the DCM modules in SPM if I have only
> the time series of ROIs? Should I constructed a SPM.mat
> file? Any suggestion would be appreciated.
>
> Thanks and best wishes,
>
> Zhi
>
>
>
>
>
>
--
Guillaume Flandin, PhD
Wellcome Centre for Human Neuroimaging
UCL Queen Square Institute of Neurology
London WC1N 3BG
|
