In article <36FA2076.73CAAB9F at psychologie.uni-wuerzburg.de>,
tietze at psychologie.uni-wuerzburg.de (Heiko Tietze) writes:
> Nice usegroup - why isn't it available on usenet ?
> Last discussed in 92, I couldnt find any link. So here again: Is there a
> way to determine artifacts with at least 99% hits ? My amplitude based
> handling detects aprox. 90% but I'm not shure about the other trials,
> which has been controlled visual, resulting in a very bad averaging. My
> idee is to use the variance of the channels by single point
> (disimilarity) and/or the deviation from SD over the whole trial for a
> single channel. The first might detect eye-blinks, the second a
> shifting. The classical method with power at spectral bands could be
> applicated too. Is this a good way and primary where can I find progs or
> source code for it ?
> Thanks and regards Heiko.
>
Peter Anderer developed a method of artifact removal based on neural
networks, if I remember correctly. The only reference I have is :
Artifact processing in topographic mapping of electroencephalographic
activity in neuropsychopharmacology. Psych. Res. 45:79-93 1992
regards,
Philippe