[Homer-users] Short-distance Channel

[Franziska Klein]

        External Email - Use Caution        

Dear homer-users,


I recorded some pilot data for the first time additionally with short-distance channels (NIRx system) and since Homer2 has the functionality to regress out the extracerebral activity from the fNIRS data I tried to use this toolbox.


I started with a very basic pipeline I reproduced from a webinar by Jonathan Perry (offered on the NIRx webpage). Unfortunately it doesn't work. For all my five recordings I receive one of the following messages in the MATLAB command window:


'Design matrix is poorly scaled...(RCond=3.044902e-15)' or 'Design matrix is VERY poorly scaled...(RCond=0.000000e+00), cannot perform computation'


I know that this is due to the design matrix but I don't understand the reason for this.  I attached the *.cfg file so that you can see the pipeline. I used the default values with the exception that I changed the rhoSD_ssThresh value to 1.0 because for some reason my probeInfo.mat file saved the coordinates in cm instead of mm and I changed flagSSmethod from 1 to 0. However, using the default for flagSSmethod results in the same message.


Is there anything I am doing wrong? What could be the reason and how can I fix this? Is it correct that the aux variable is a vector (length of timepoints) consisting only of ones? In the archive of the mailing list I found a similar problem and when I understand it correctly the problem was an empty 's' matrix but this does not apply in my case.


I would appreciate any help.

Thanks in advance,

Franziska



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/homer-users/attachments/20180822/8b53e1d3/attachment.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Short_channel.cfg
Type: application/octet-stream
Size: 672 bytes
Desc: Short_channel.cfg
Url : http://mail.nmr.mgh.harvard.edu/pipermail/homer-users/attachments/20180822/8b53e1d3/attachment.obj