[Eeglablist] how to apply ICA methods on pre-clinical electrode configurations

Mon Jan 18 06:36:59 PST 2021

I found a public source for a 16 channel locations file and created the EEGlab dataset using my mouse data, the channel locations file and the ‘w’ and ‘s’ variable outputs from ICA… Looks like I am in business.

STUDY = []; CURRENTSTUDY = 0; ALLEEG = []; EEG=[]; CURRENTSET=[];
EEG = pop_importdata('dataformat','array','nbchan',16,'data','data','setname','Whisk','srate',1221,'pnts',0,'xmin',0,'chanlocs','E:\\electrode_positions_16channel.sfp','icaweights','w','icasphere','s');
[ALLEEG EEG CURRENTSET] = pop_newset(ALLEEG, EEG, 0,'gui','off');
eeglab redraw
pop_prop( EEG, 0, [1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16], NaN, {'freqrange' [2 80] });

However, I should still dig into the code (where ICA weights are applied) and learn these methods

Cheers,
Brian

From: Scott Makeig <smakeig at gmail.com>
Sent: Monday, January 18, 2021 8:43 AM
To: Brian Harvey <brian.harvey at biogen.com>
Cc: eeglablist at sccn.ucsd.edu
Subject: Re: [Eeglablist] how to apply ICA methods on pre-clinical electrode configurations

________________________________
*** EXTERNAL EMAIL: Use caution before replying, clicking links, and opening attachments ***
________________________________
The spatial (channel) weights would quantify the degree to which a given dynamic is truly 'population level' or not ...

On Sun, Jan 17, 2021 at 7:07 PM Brian Harvey <brian.harvey at biogen.com<mailto:brian.harvey at biogen.com>> wrote:
Thanks again Dr. Makeig,

Honestly, I don't really care about the "where" as the probe is entirely (almost) in sensory thalamus ( 16 channels 50um separation distance). Each channel is picking up LFP but also spiking activity (sample rate is ~24K). I am detecting different spiking activities at channels in response to the stimulus but am not so much concerned about the location but isolating the population activity that represents the 10Hz afferent input. My goal in this attempt is to avoid a biased selection of channels for quantification.

Best,
Brian

________________________________
From: Scott Makeig <smakeig at gmail.com<mailto:smakeig at gmail.com>>
Sent: Sunday, January 17, 2021 6:55 PM
To: Brian Harvey <brian.harvey at biogen.com<mailto:brian.harvey at biogen.com>>
Cc: eeglablist at sccn.ucsd.edu<mailto:eeglablist at sccn.ucsd.edu> <eeglablist at sccn.ucsd.edu<mailto:eeglablist at sccn.ucsd.edu>>
Subject: Re: [Eeglablist] how to apply ICA methods on pre-clinical electrode configurations

________________________________
*** EXTERNAL EMAIL: Use caution before replying, clicking links, and opening attachments ***
________________________________
Well, you do want to know & see which electrode combination each IC projects to, no?  You need to make an EEG (dataset) structure - then all the EEGLAB tools will be readily available...

Scott

On Sun, Jan 17, 2021 at 6:48 PM Brian Harvey <brian.harvey at biogen.com<mailto:brian.harvey at biogen.com>> wrote:
Thanks Dr Makeig for your response,

What if one only wanted to see the spectra of each component?  ICA doesn't require info about channel locations and in this application, I am not concerned about topographies (thus topoplot.m is moot)...

I am naive in regards how to manually use the w and s variables from the output of ICA to manually generate the PSDs of each component.

More specifically, the experiment is aimed at evaluating thalamic activity in response to whisker deflection (10Hz)... I am hoping to find a component that can isolate the driven neural activity... I can do this in the channel space using FFT but was thinking to apply ICA and ignore channels....

Cheers,
Brian

________________________________
From: Scott Makeig <smakeig at gmail.com<mailto:smakeig at gmail.com>>
Sent: Sunday, January 17, 2021 6:27 PM
To: Brian Harvey <brian.harvey at biogen.com<mailto:brian.harvey at biogen.com>>
Cc: eeglablist at sccn.ucsd.edu<mailto:eeglablist at sccn.ucsd.edu> <eeglablist at sccn.ucsd.edu<mailto:eeglablist at sccn.ucsd.edu>>
Subject: Re: [Eeglablist] how to apply ICA methods on pre-clinical electrode configurations

________________________________
*** EXTERNAL EMAIL: Use caution before replying, clicking links, and opening attachments ***
________________________________
Brian -

The commandline function topoplot.m has a mode in which you can plot interpolated electrode positions specified in a rectangular grid (with arbitrary 'missing nodes', allowing any grid shape). This can be used to plot the IC 'maps'.

As for your main question, read the EEGLAB wiki on building an EEG dataset.

Scott

On Sat, Jan 16, 2021 at 11:30 PM Brian Harvey <brian.harvey at biogen.com<mailto:brian.harvey at biogen.com>> wrote:
Hi all,

A bit of a hack attempt here... I have LFP data using a 16 channel linear probe in VPL/VPM thalamus (in mouse)... I can successfully run ICA using CUDAICA

[w,s] = cudaica(LFP,'extended', 1, 'maxsteps', 2048);

If I am only interested in the spectral properties of the components (not scalp topographies etc) how can I manually plot them for rejection or selection? EEGlab functions require channel locations to proceed further and I am uncertain how to hack.

Thanks for any insight/assistance

Brian

_______________________________________________
Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html<https://urldefense.proofpoint.com/v2/url?u=http-3A__sccn.ucsd.edu_eeglab_eeglabmail.html&d=DwMFaQ&c=n7UHtw8cUfEZZQ61ciL2BA&r=TRx_2zPEKjt2eUqOrrQ2qy2yNPA3tLhBWOLba81oTK8&m=r-h6S6KfqT9rZVFIbDIP4Qt9PJOt7DgkUF_y33pqIFU&s=hd5WnEnpMPbuYe2mRVR_DdqIuEBXHGBxmflGX0E1SE4&e=>
To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu<mailto:eeglablist-unsubscribe at sccn.ucsd.edu>
For digest mode, send an email with the subject "set digest mime" to eeglablist-request at sccn.ucsd.edu<mailto:eeglablist-request at sccn.ucsd.edu>

--
Scott Makeig, Research Scientist and Director, Swartz Center for Computational Neuroscience, Institute for Neural Computation, University of California San Diego, La Jolla CA 92093-0559, http://sccn.ucsd.edu/~scott<https://urldefense.proofpoint.com/v2/url?u=http-3A__sccn.ucsd.edu_-257Escott&d=DwMFaQ&c=n7UHtw8cUfEZZQ61ciL2BA&r=TRx_2zPEKjt2eUqOrrQ2qy2yNPA3tLhBWOLba81oTK8&m=r-h6S6KfqT9rZVFIbDIP4Qt9PJOt7DgkUF_y33pqIFU&s=bdcVmA8PJfQcuHArK8GRazNFPSmeLjEAzUay6q83IzI&e=>

--
Scott Makeig, Research Scientist and Director, Swartz Center for Computational Neuroscience, Institute for Neural Computation, University of California San Diego, La Jolla CA 92093-0559, http://sccn.ucsd.edu/~scott<https://urldefense.proofpoint.com/v2/url?u=http-3A__sccn.ucsd.edu_-257Escott&d=DwMFaQ&c=n7UHtw8cUfEZZQ61ciL2BA&r=TRx_2zPEKjt2eUqOrrQ2qy2yNPA3tLhBWOLba81oTK8&m=ANRoi0C-gI8-ChG_j2Zdol0zbtjdmrVTVA5zvyfQosY&s=ZZ8IlHg8YD5fI9VNcC5Y4mrLqWKh4oW4ZQKpoRBxZ0I&e=>

--
Scott Makeig, Research Scientist and Director, Swartz Center for Computational Neuroscience, Institute for Neural Computation, University of California San Diego, La Jolla CA 92093-0559, http://sccn.ucsd.edu/~scott<https://urldefense.proofpoint.com/v2/url?u=http-3A__sccn.ucsd.edu_-257Escott&d=DwMFaQ&c=n7UHtw8cUfEZZQ61ciL2BA&r=TRx_2zPEKjt2eUqOrrQ2qy2yNPA3tLhBWOLba81oTK8&m=97SP4D1uyPHl-F4Qf1xb_9b4QdiZRT4y-9jscrdZKuw&s=y5ad-KnCYLUNDlQAw9olXjAhdVLeJc7WlCwj_CQOMRE&e=>