[Eeglablist] How to read large neuroscan dataset in eeglab?

[arno delorme]

Dear Xiaoyun,

there is no obvious solution to this problem. In the latest EEGLAB  
version (currently only in house) we have memory mapping which means  
that the data stays on disk. However, even in this version you would  
not be able to import your dataset because Matlab has first to import  
the full dataset in memory and then to save it. The trick would be to  
rewrite part of the loadcnt.m function so that the data is rewritten  
on disk in a standard float file as it is being read (float32 format).

Unfortunately unless someone is motivated to undertake this, it will  
not be implemented. It is a matter of a couple hours of programming  
and I can guide someone through this if anybody is interested. If you  
do it, I can send you the latest in house EEGLAB version so you can  
test it further.

Also, it would be good to have this for other popular data formats.
Best,

Arno

On 18 nov. 08, at 11:44, <liangxiaoyunseu at hotmail.com> <liangxiaoyunseu at hotmail.com 
 > wrote:

> Hi all,
>        I was trying to use the eeglab to process the eeg dataset  
> acquired with fMRI simultaeously using neuroscan EEG system.  
> However, the dataset was very large (more than 1G). Even if I used  
> 64 bit matlab, I still found the same problem of "out of memory".  
> Did anyone come across the same problem before.
> Do you guys have any good ideas to address such issues? In addition,  
> I learned that there is a method to read just some channels or  
> epochs once. But I don't know how to deal with it. If anyone could  
> tell me about it, I would greatly appreciate that!
>        Thanks a lot!
>
>
>        Best,
>        Xiaoyun Liang
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20081124/55cc1459/attachment.html>