Re: [Faster-eeg-list] Some observations and suggestions

[Grega R. <gre...@ps...>]

Dear Hugh,

I do find FASTER useful, I especially like the simplicity of use and one step preprocessing it affords. I would however like to know how it compares to other preprocessing options. For that reason, we're running data that we have collected in an oddball paradigm (128 channels) and preprocessed "by hand" through both FASTER and EP toolbox. I'd like to see how data preprocessed by an expert differs from fully automatic preprocessing. Previously working primarily with the fMRI, I'm used to the latter and it would be great to have reliable fully automatic preprocessing available for EEG as well.

In regards to binica, I'm running analyses on an Apple Mac Server (with Mac OS X Leopard and the latest version of MATLAB). I have:

1/ installed the version of binica compiled for 32/64 bit Intel Apple hardware, 
2/ set the path to binica folder in Matlab,
3/ set ICABINARY in eeglab/functions/sigprocfunc/icadefs.m to the correct path
4/ changed the lines following % Do ICA % in FASTER_process.m to specify 'icatype', 'binica' in calls to pop_runica

I did no other changes. I am assuming it might even work if 'icatype', 'binica' is omitted from the call as pop_runica should call binica by default and then switch to runica if it can not find / run binica.

Additionally, there was a conflict with fieldrip so I removed fieldtrip paths when running FASTER and it all runs fine now.

All the best,

Grega



On Oct 14, 2010, at 6:55 PM, Hugh Nolan wrote:

> Dear Grega,
> thank you very much for your comments!
> 
> In response to your points:
> 
>>> Bugs
>>> - When running a job on a folder with multiple .set and .fdt files for a study, FASTER creates a folder for each of the subjects, however it only copies .set files and not the corresponding .fdt files, which leads to an error when .fdt files are not found in the next step, but have to be moved there manually before restarting the job.
> Thank you for this, there are unfortunately 2 versions of a particular
> file due to a MATLAB version oddity - I forgot to copy the changes
> from one into the other! A new version of FASTER will be available in
> the next day or two with this and a few other bug fixes, as well as a
> couple of new features. If you wish to try the in-progress version
> (with the bug fixes, but not all the new features), email me
> personally (no...@tc...).
> 
>>> - When saving job, the job is always saved to the existing file and there is no option for a "Save As" operation.
> There is a "Save As" option in the "More Options" menu (top left of
> the GUI). Maybe it would be a good idea to rename this "File Options"
> to make it more obvious?
> 
>>> Questions
>>> - It is not entirely clear what happens when the .set files are already epoched. Is the existing epoching used, or is the whole recording considered a single epoch?
> Existing epoching is used, and epoch removal / rejection can be
> performed on these datasets. If markers are entered into the "Epoching
> markers" box, further epoching based on these markers will take place.
> I think I will add an explicit "Do Markered Epoching" option to try
> and clear up any ambiguity.
> 
>>> Suggestions
>>> - At this time the use of runica is "hard coded" into the script. To use binica instead I had to change the script itself. It would be nice to have that as an option under ICA options.
> Would you so kind as to send me (no...@tc...) or Rob
> (rob...@tc...) this change? I haven't had much success with
> binica myself, but if you have got it working with FASTER, I would be
> happy to integrate it!
> 
>>> - It would be great if one had the opportunity to review, which were the ICA components found and which of them were rejected. One possibility would be to save the mixing matrices in the "Intermediate" folder. An even user-friendlier option would be to generate component topography images and save them as a PNG image or a PDF document
> This is a very good idea - I'll include it in the update.
> 
> I hope you are finding FASTER useful.
> 
> Hugh
> 
> --
> Hugh Nolan,
> Trinity Centre for Bioengineering,
> Printing House,
> Trinity College Dublin.
> 
> Tel: +353861297722
> 
> ------------------------------------------------------------------------------
> Beautiful is writing same markup. Internet Explorer 9 supports
> standards for HTML5, CSS3, SVG 1.1,  ECMAScript5, and DOM L2 & L3.
> Spend less time writing and  rewriting code and more time creating great
> experiences on the web. Be a part of the beta today.
> http://p.sf.net/sfu/beautyoftheweb
> _______________________________________________
> Faster-eeg-list mailing list
> Fas...@li...
> https://lists.sourceforge.net/lists/listinfo/faster-eeg-list
> 




-- 
Asist. Prof. Grega Repovš, Ph.D.
Department of Psychology
University of Ljubljana
Aškerčeva 2
SI-1000 Ljubljana
tel: +386 1 241 1175
email: gre...@ps...