<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/loose.dtd">
<html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body style="font-family:Arial;font-size:14px"><p>
<br><br><br><br>
Dear Tim,<br><br>
thank you very much for all your emails and advises. I answer in detail to your following email:<br><br>
Quoting Tim Mullen <<a href="mailto:mullen.tim@gmail.com">mullen.tim@gmail.com</a>>:</p>
<blockquote style="border-left:2px solid blue;margin-left:2px;padding-left:12px;" type="cite">
<div dir="ltr">
Dear Vito,
<div>
</div>
<div>
Correct, SIFT can work on both continuous data or epoched multi-trial data and uses the third dimension (epochs) to determine which is the case. If you wish to operate on epoched single trials, you may concatenate the trials into a 2D matrix (chs x pnts*trials) In this case, bear in mind that if you're using a sliding window approach, you will need to discard the results within the period corresponding to 1/2 of the window length at the beginning and end of each trial, since the sliding window will have overlapped two trials here. </div>
<div>
<br><br>
</div>
</div>
</blockquote>
I have a big problem with Granger Analysis (also using GCCA or BSMART) because my data are not really easy to analyze. I record local field potential from several brain regions of neonatal rats. I am analyzing the casual flow from the Hippocampus (HIP) to the Prefrontal Cortex (PFC). This flow should be unidirectional because the anatomic connections are just in one direction. So my goal is to proof, using information flow (Granger, Transder Entropy, etc.) that I have a direct connection (just monosinaptic) from HIP to PFC. Obviously I have also a path from PFC to HIP but this path is not direct and the information flow takes more time. First of all I don't know if I can proof this kind of connectivity using Granger, I would expect that the flow from HIP to PFC (where I have direct connections) is higher than the flow from PFC to HIP (where I don't have direct connections).<br><br>
The signals are not easy to analyze because:<br><br>
1) In this case we just measure the "spontaneous" activity that means that we do not stimulate the animal and he is sleeping during the recording (we use urethane anesthesia). So I don't have "naturally epoched" data and I window the data (sliding window) also to improve the stationarity. I have a window of 500ms and I use two approaches:<br>
a) I consider each window as my entire trace and I calculate a MVAR model for each window<br>
b) I divide each window in 10 subwindows of 50ms and I consider each subwindow as an epoche (as a realizzation of the same random process)<br>
This is why asked about <br>
I found impossible to have one model for the entire recording (my recording lasts 40 min). The best compromise beetween consistence and stability (using Anil Seth toolbox) is to consider a maximum lenght of the trace of 500ms. With SIFT I consider the entire recording and this is probably why I don't have stability and consistence. As I have understood fine SIFT create one model for the entire trace (of 45 min) as is normal to do (but not in my case I think). <br><br>
2) In the neonatal rats the activity is highly discontinuous and I have bursts of activity interleaved by noise (the noise level is my baseline and I have 5-6 bursts per minute of 0.5-8 sec each). In this way the signal is very far to be stationary, also because the activity (the bursts of activity) is very not stationary.<br><br>
In this situation it is very difficult to have stationarity, linearity, good model consistence, white residuals, etc... In addition also when the criteria are satisfied (in few parts of the signal trace) I cannot find high flow density. Also Steven Bressler that was here in December and gave a look to my data told me that the problem is difficult. Probably a linear Granger approach is not correct for this data.<br><br>
Anyway I developed this method that gives some results. The method is based on calculating a different model for each window during the sliding and on the time shifting of the time series, as I wrote before.<br><br>
You wrote: "For single-trial analysis, unless the number of variables (channels/sources) you are modeling is low, you will likely need to use regularized VAR estimators (ridge regression or sparse VAR) or a Kalman filter approach (est_fitMVARKalman) to achieve a reasonable model fit."<br><br>
I have just 2 channels (HIP and PFC). I can also have a third region that I guess could be an intermidiate region for the flow from PFC to HIP (the flow from HIP to PFC should be direct as I worte before).<br>
Sorry, Tim, I don't know these methods: "ridge regression or sparse VAR) or a Kalman filter approach (est_fitMVARKalman)". Where can I find some explanation about these methods?<br>
I just tried to differenciate but I don't have better results.<br><br>
You wrote:<br>
It appears to me that your data still has some stationarity or other issues leading to model instability. I recommend trying the following:<br><br>
1) clean your data to remove trials, channels/sources, or data subspaces with large artifacts -- especially blinks and muscle artifacts, which are not well-modeled by a stationarity VAR process<br>
2) downsample the data to an appropriate sampling rate (256 Hz is typical for EEG)<br>
3) apply local detrending in SIFT pre-processing, if drift is present<br>
4) decrease the window length to (e.g.) 0.5 sec to produce a more locally stationarity signal.<br>
5) increase the model order if needed (you might check the model order selection procedure to help determine this)<br><br>
I don't have artifacts (this is one of the few good things!) but the signal often seem to contain artifacts because the local field potential activity has often very big peaks (this a big difference beetween local field potential and EEG or fMRI).<br>
If I downsample to 800 Hz, if I downsample more the consistence is very low and also the stationarity is worse.<br>
I apply everytime detrending and the model order changes window by window but the average is around 10.<br><br>
If you have any suggestion I would appreciate a lot!<br><br>
Vito<br><blockquote style="border-left:2px solid blue;margin-left:2px;padding-left:12px;" type="cite">
<div dir="ltr">
<div>
<br><br><br><br>
</div>
<div>
Tim</div>
<div class="gmail_extra">
<br><br><div class="gmail_quote">
On Sun, Jan 19, 2014 at 12:08 PM, Vito de Feo <span dir="ltr"><<a href="mailto:vito.defeo@zmnh.uni-hamburg.de" target="_blank">vito.defeo@zmnh.uni-hamburg.de</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div style="font-family:Arial;font-size:14px">
<p>
<u>Dear all,<br><br><br>
sorry if I write again. I am very interested to know better SIFT!<br><br>
I have understood that the stability test includes also the stationarity test. So sorry for my prevoius question.<br>
Now I am trying to understand if SIFT uses a multitrial approch or a single trial approach. I guess that if the signal is not epoched it use a single trial approach.<br><br>
Now I have problems with the consistency (I attach two pictures). The consistency is very low as you can see. What can I do? Should I increase the model order or should I decrease the moving windows length? (now it is 1.5 s, I could decrese to 0.5 s beacuse the signal is strongly not stationary).<br><br>
Thank you!</u></p>
<div>
<br><br><u>Vito <br><br>
Quoting Andreas Widmann <<a href="mailto:widmann@uni-leipzig.de" target="_blank">widmann@uni-leipzig.de</a>>:</u></div>
<p>
</p>
<div>
<div>
<blockquote style="border-left:2px solid blue;margin-left:2px;padding-left:12px" type="cite">
<div>
<u>Dear all,</u></div>
<div>
<u> </u></div>
<div>
<u>not directly related to your question and SIFT, but eegfilt is deprecated and I would recommend not using it any longer.</u></div>
<div>
<u> </u></div>
<div>
<u>Best,</u></div>
<div>
<u>Andreas</u></div>
<div>
<br><u>Am 18.01.2014 um 15:47 schrieb "jfochoaster ." <<a href="mailto:jfochoaster@gmail.com" target="_blank">jfochoaster@gmail.com</a>>:<br>
</u></div>
<blockquote type="cite">
<div>
<div dir="ltr">
<div>
<div>
<div>
<div>
<u>Hello all,<br>
</u></div>
<u>I'm following the SIFT tutorial, the section 6.5.1.3 is about filtering, talk about eegfilt, about the zero-phase (acausal) filter<br>
</u></div>
<u>Is better forget this section of filtering and use the recommendations in the past emails?<br>
</u></div>
<div>
<u>Are these recommendation critical for the analysis?, I mean, there is a lot of work about MVAR models in ECoG data</u></div>
<div>
<u> </u></div>
<u>Best wishes<br>
</u></div>
<u>John</u></div>
<div class="gmail_extra">
<br><br><div class="gmail_quote">
<u>On Fri, Jan 17, 2014 at 11:05 PM, <a href="mailto:mullen.tim@gmail.com" target="_blank">mullen.tim@gmail.com</a> <span dir="ltr"><<a href="mailto:mullen.tim@gmail.com" target="_blank">mullen.tim@gmail.com</a>></span> wrote:</u><br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<u>Oh thats interesting. I had not seen Anil's multitaper filter (might be fairly recent). But possibly it is exactly the same approach that is in Cleanline. If this is the method advocated by Mitra and Pesaran as in the Chronux toolbox then indeed its the same. And highly recommended.</u><br><div>
<u>-----Original Message-----<br>
Date: Friday, January 17, 2014 1:21:30 pm<br>
To: <a href="mailto:mullen.tim@gmail.com" target="_blank">mullen.tim@gmail.com</a><br>
Cc: <a href="mailto:trotta_gabriele@yahoo.com" target="_blank">trotta_gabriele@yahoo.com</a>, <a href="mailto:drcoben@gmail.com" target="_blank">drcoben@gmail.com</a>, <a href="mailto:mmiyakoshi@ucsd.edu" target="_blank">mmiyakoshi@ucsd.edu</a>, <a href="mailto:widmann@uni-leipzig.de" target="_blank">widmann@uni-leipzig.de</a>, <a href="mailto:eeglablist@sccn.ucsd.edu" target="_blank">eeglablist@sccn.ucsd.edu</a><br>
From: "Vito De Feo" <<a href="mailto:vito.defeo@zmnh.uni-hamburg.de" target="_blank">vito.defeo@zmnh.uni-hamburg.de</a>><br>
Subject: Re: [Eeglablist] Filter causality pop_eegfiltnew<br>
</u></div>
<u>Before using the Cleanline (that I used today for the first time) I did't use the notch filter, I used a multi taper filtering made by Anil Seth. I know that filtering is very bad for later VAR modeling, especially notch and high pass. Low pass is better (usually I use multi taper filtering to remove the noise lines and a low pass causal filter with cut off filtering of 100 Hz).<br>
Do you think is ok Tim?<br>
Best<br>
Vito<br><br><br>
Il giorno 17/gen/2014, alle ore 20:53, <a href="mailto:mullen.tim@gmail.com" target="_blank">mullen.tim@gmail.com</a> ha scritto:</u><br><div>
<br><u>> Do not notch filter your data! This can be very bad for later VAR modeling -- and IMO bad in general. You can use an adaptive spectral regression method such as that in the Cleanline plugin for eeglab to remove line noise.<br>
><br>
> See Barnett and Seth 2011 and Mitra and Pesaran 1999 for theoretical discussions.<br>
><br>
> Rob, there is no video of the SIFT workshop but the lecture pdfs are online at the eeglab workshop page.<br>
><br>
> Tim<br>
> -----Original Message-----<br>
> Date: Friday, January 17, 2014 10:18:32 am<br>
> To: "<br>
</u></div>
<div>
<div>
<u>_______________________________________________<br>
Eeglablist page: <a href="http://sccn.ucsd.edu/eeglab/eeglabmail.html" target="_blank">http://sccn.ucsd.edu/eeglab/eeglabmail.html</a><br>
To unsubscribe, send an empty email to <a href="mailto:eeglablist-unsubscribe@sccn.ucsd.edu" target="_blank">eeglablist-unsubscribe@sccn.ucsd.edu</a><br>
For digest mode, send an email with the subject "set digest mime" to <a href="mailto:eeglablist-request@sccn.ucsd.edu" target="_blank">eeglablist-request@sccn.ucsd.edu</a></u></div>
</div>
</blockquote>
</div>
<br><br clear="all"><br><u>--<br>
John Ochoa<br>
Docente de Bioingeniería<br>
Universidad de Antioquia</u></div>
</div>
</blockquote>
</blockquote>
<br><br><br>
</div>
</div>
<div>
<u>--<br>
Pflichtangaben gemäß Gesetz über elektronische Handelsregister und Genossenschaftsregister sowie das Unternehmensregister (EHUG):<br><br>
Universitätsklinikum Hamburg-Eppendorf<br>
Körperschaft des öffentlichen Rechts<br>
Gerichtsstand: Hamburg<br><br>
Vorstandsmitglieder:<br>
Prof. Dr. Christian Gerloff (Vertreter des Vorsitzenden)<br>
Prof. Dr. Dr. Uwe Koch-Gromus<br>
Joachim Prölß<br>
Rainer Schoppik</u></div>
</div>
</blockquote>
</div>
<br><br clear="all"><div>
</div>
<u>--<br>
--------- αντίληψη ----------- </u></div>
</div>
</blockquote>
<br><br><br><br>-- <br>Pflichtangaben gemäß Gesetz über elektronische Handelsregister und
Genossenschaftsregister sowie das Unternehmensregister (EHUG):<br><br>
Universitätsklinikum Hamburg-Eppendorf<br>
Körperschaft des öffentlichen Rechts<br>
Gerichtsstand: Hamburg<br><br>Vorstandsmitglieder:<br>
Prof. Dr. Christian Gerloff (Vertreter des Vorsitzenden)<br>
Prof. Dr. Dr. Uwe Koch-Gromus<br>
Joachim Prölß<br>
Rainer Schoppik<br>
</body></html>