hi, Christian,

I still have a question of doing post-stats in MELODIC. I have tried your suggestions of how to do mixture modelling after ICA estimation is done. So by using the following call:

melodic -i <bgimage> --ICs=melodic_IC --mix=melodic_mix [ --smode=melodic_Smode ] [--Tdes=design.mat --Tcon=design.con ] -v --report [ --Ostats ]

I will get the thresholded spatial maps of the ICs (posterior probability p<0.5), and after clicking the maps, the raw Z-maps and probability maps will appear.

I am wondering how melodic does the Z-transformation based on the input "melodic_IC" and "melodic_mix", because it needs the standard deviation of the  background noise to do the transformation, right? Or the information of the residual noise is alreay included in the input? Since this call doesn't do PICA estimation from the start, I am not sure about the noise estimation part. I guess I am missing some point here ...

Thank you very much for any advice!

Best Regards,

Yanmei

Christian F. Beckmann wrote:

Hi

yes, you can call the mixture model fitting and report generation using the command line version of melodic

melodic -i <bgimage> --ICs=melodic_IC --mix=melodic_mix [ --smode=melodic_Smode ] [--Tdes=design.mat --Tcon=design.con ] -v --report [ --Ostats ]

<bgimage> needs to be in the space of the IC maps in melodic_IC and the optional smode adds subject-modes to the report. The --Tdes and --Tcon options tell melodic which GLM design to use for post-hoc OLS fitting - you can do the same for GLM fitting to the smdes, simply add the --Sdes= and --Scon= options . Finally, the --Ostats option makes sure that the mixture model probability maps get saved.

hope this helps

Christian

_______________________________________________

Christian F. Beckmann, DPhil

Senior Lecturer, Clinical Neuroscience Department

Division of Neuroscience and Mental Health

Imperial College London

Hammersmith Campus, Rm 247, Cyclotron Bldg

Du Cane Road, London W12 0NN, UK

Tel.: +44 (0) 208 383 8598   ---   Fax: +44 (0) 208 383 2029

Email: [log in to unmask] <mailto:[log in to unmask]>   http://www.imperial.ac.uk/medicine/people/c.beckmann/

On 18 Sep 2007, at 18:52, Yanmei Tie, Ph.D. wrote:

Hi Steve,

Sorry for the confustion. I mean, is there a way of ONLY performing the post-stats? Like, I have IC z-maps already, and just want to do mixture modeling, and/or post-hoc GLM analyses of IC time courses against a GLM design matrix ...

Thank you very much!

yanmei

Steve Smith wrote:

Hi - I'm a little confused as this is exactly the default behaviour  that you get when you run MELODIC from the GUI.

Cheers.

On 18 Sep 2007, at 14:49, Yanmei Tie wrote:

hi all,

I wonder if there is a way that I can perform post-stats (Gaussian/ Gamma

mixture modeling and alternative hypothesis test with thresholding)  using IC

z-maps as input. Is there an option in melodic command-line?

Thanks very much for any suggestions.

Best regards,

yanmei

------------------------------------------------------------------------ ---

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] <mailto:[log in to unmask]>    http://www.fmrib.ox.ac.uk/~steve

------------------------------------------------------------------------ ---