[Eeglablist] ASR – include EOG/ECG channels, apply before referencing?

[Makoto Miyakoshi]

> Awesome, thanks! Is there any advantage to using clean_rawdata's built-in
filter vs. ERPlab filter that automatically does DC mean subtraction?

I have never used ERPlab so hard to say anything. I know clean_rawdata's
filter is IIR with no phase shift though.

> Also, when you say you get better ASR results with SD=20, do you mean for
burst criterion or also line noise criterion?
burst criterion. You don't need to use line noise criterion if your data is
clean. Enter -1 to disable it.

Makoto

On Thu, May 7, 2015 at 12:21 PM, Kevin Tan <kevintan at cmu.edu> wrote:

> Awesome, thanks! Is there any advantage to using clean_rawdata's built-in
> filter vs. ERPlab filter that automatically does DC mean subtraction?
>
> Also, when you say you get better ASR results with SD=20, do you mean for
> burst criterion or also line noise criterion?
>
> Thanks again, this is really helping me a lot!!
>
> –Kevin
> --
> Sent with Gmail for Android
> On May 7, 2015 2:27 PM, "Makoto Miyakoshi" <mmiyakoshi at ucsd.edu> wrote:
>
>> Dear Kevin,
>>
>> > Should I remove DC mean before running ASR?
>>
>> Yes.
>> That is why the high-pass filter process is positioned at the first step
>> in the clean_rawdata(). If DC and slow wave (< 1.0 Hz) are not removed, ASR
>> does not work well.
>>
>> Makoto
>>
>> On Tue, May 5, 2015 at 6:14 PM, Kevin Tan <kevintan at cmu.edu> wrote:
>>
>>> Thanks so much for the response! Much appreciated.
>>>
>>> I have another question: the BioSemi runs off a battery; "DC mean" has
>>> to be removed until the EEG signal looks normal.
>>>
>>> Should I remove DC mean before running ASR? Seems to make a difference
>>> in terms of # of blocks ASR cleans.
>>>
>>> Cheers,
>>> Kevin
>>>
>>> --
>>> Kevin Alastair M. Tan
>>> Lab Manager/Research Assistant
>>> Department of Psychology & Center for the Neural Basis of Cognition
>>> Carnegie Mellon University
>>>
>>> Baker Hall 434
>>> <https://www.google.com/maps/place/40%C2%B026%2729.5%22N+79%C2%B056%2744.0%22W/@40.4414869,-79.9455701,61m/data=!3m1!1e3!4m2!3m1!1s0x0:0x0>
>>>  | kevintan at cmu.edu | tarrlab.org/kevintan
>>> <http://tarrlabwiki.cnbc.cmu.edu/index.php/KevinTan>
>>>
>>> On Tue, May 5, 2015 at 1:43 PM, Makoto Miyakoshi <mmiyakoshi at ucsd.edu>
>>> wrote:
>>>
>>>> Dear Kevin,
>>>>
>>>> > I'm wondering whether to include EOG and ECG channels in ASR like in
>>>> traditional ICA? Also, whether to apply ASR before referencing?
>>>>
>>>> Yes, for the purpose of consistency. If I understand correctly (I say
>>>> this because the main algorithm is totally Christian's, and I just wrote a
>>>> wrapper) ASR computes correlation matrix of the channels. You may want to
>>>> include all channels rather than separate some of them (unless you reject
>>>> them permanently). Applying it before rereferencing makes sense, since you
>>>> want to clean the data (i.e. exclude noisy channels) before referencing.
>>>>
>>>> Your pipeline looks fine to me. Good luck!
>>>>
>>>> Makoto
>>>>
>>>> On Fri, May 1, 2015 at 1:36 PM, Kevin Tan <kevintan at cmu.edu> wrote:
>>>>
>>>>> Hello Dr. Miyakoshi & EEGlab list members,
>>>>>
>>>>> I'm exploring artifact subspace reconstruction (clean_rawdata plugin)
>>>>> as an alternative to threshold-based artifact rejection.
>>>>>
>>>>> I'm wondering whether to include EOG and ECG channels in ASR like in
>>>>> traditional ICA? Also, whether to apply ASR before referencing?
>>>>>
>>>>> I am using a 128-channel BioSemi ActiveTwo with 8 external channels:
>>>>> M1, M2, bipolar hEOG, bipolar left vEOG, bottom right vEOG, and ECG.
>>>>>
>>>>> BioSemi always records unreferenced. I recorded at 512hz for the sake
>>>>> of comp resources.
>>>>>
>>>>> Here's my preliminary preproc pipeline:
>>>>>
>>>>>    1. Import data using BDF plugin (unreferenced)
>>>>>    2. Add channel location info
>>>>>    3. CleanLine
>>>>>    4. ASR (including EOG/ECG?)
>>>>>    5. Reference to 128ch scalp average
>>>>>    6. ERPlab:
>>>>>    1. Create Eventlist
>>>>>       2. Extract event-based epochs
>>>>>       3. Create averaged ERPs
>>>>>       4. Lo-pass @ 30hz
>>>>>
>>>>> Any advice would be appreciated :-)
>>>>>
>>>>> Many thanks,
>>>>> Kevin
>>>>> --
>>>>> Kevin Alastair M. Tan
>>>>> Lab Manager/Research Assistant
>>>>> Department of Psychology & Center for the Neural Basis of Cognition
>>>>> Carnegie Mellon University
>>>>>
>>>>> Baker Hall 434
>>>>> <https://www.google.com/maps/place/40%C2%B026%2729.5%22N+79%C2%B056%2744.0%22W/@40.4414869,-79.9455701,61m/data=!3m1!1e3!4m2!3m1!1s0x0:0x0>
>>>>>  | kevintan at cmu.edu | tarrlab.org/kevintan
>>>>> <http://tarrlabwiki.cnbc.cmu.edu/index.php/KevinTan>
>>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> Makoto Miyakoshi
>>>> Swartz Center for Computational Neuroscience
>>>> Institute for Neural Computation, University of California San Diego
>>>>
>>>
>>>
>>
>>
>> --
>> Makoto Miyakoshi
>> Swartz Center for Computational Neuroscience
>> Institute for Neural Computation, University of California San Diego
>>
>


-- 
Makoto Miyakoshi
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20150507/83d62910/attachment.html>