[Eeglablist] double event markers from Biosemi .bdf files

[Andre Achim]

The duration of the pulse code is directly responsible for this.  Whenever
the pulse spans two sampling bins, two consecutive identical codes appear in
the event channel.  There are also problems with codes that are emitted in
close succession.  If the signal is decimated and two codes were presented
in close succession, then the code stored in the Biosemi file is the logical
AND of the bits that were ON in any of the bins that are merged together.
The results depend on the exact timing of the bins with respect to the
duration of the emitted codes.

 

My solution, when reading the sequence of coded events is to remove codes
that (a) are identical to the previous code and (b) follow it by just one
time step.

 

If codes must be sent in close succession, e.g. code 254 (Pause OFF) to
start the acquisition and a condition code, I now make sure to insert a
reasonable delay between the two (say 5 ms).  In our first Biosemi
application, with decimation, there was no delay and we sometimes had three
different codes at three consecutive times, the first was 254, the last was
our event code and the second was the logical bit-AND of the two.  The
middle code had to be replaced by the third (event code) and the third code
deleted (since the event code appeared during the second time bin).  In a
few less favourable cases, only code 254 and the logical AND were present,
such that the event code could not be identified unambiguously and was
therefore lost.

 

An important consideration, also, in sending codes to the Biosemi system, is
to wait 1 s after the last useful part of the recording, before sending out
code 255 (Pause ON), because any 1 s epoch currently incomplete when this
code is received is not saved (except perhaps in the very latest version of
Actiview).

 

Andre Achim

 

-----Message d'origine-----
De : eeglablist-bounces at sccn.ucsd.edu
[mailto:eeglablist-bounces at sccn.ucsd.edu] De la part de Jim Kroger
Envoyé : 14 février 2005 23:14
À : eeglablist
Objet : [Eeglablist] double event markers from Biosemi .bdf files

 

Hello, I am wondering if anybody else has experienced this problem. We use a
Biosemi 128-channel.
Channel 129 has event markers. 

When using this channel to mark event times in EEGLAB, we have some kind of
ring or artifact that causes EEGLAB to mark two events instead of one. We
also try loading this into EMSE, and do not have the problem.

We're having a bit of trouble understanding why this happens. Although we
can imagine ways to program around the problem, it would be best if we
understood it.

Many thanks for any pointers.

Jim




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20050215/a402d674/attachment.html>