« thresholdDetect2 | Software | brainhull »

newDs2

newDs2 is a version of newDs that can save virtual channels without the MEG channels, and which has a more convenient filtering option.

Download and Installation

Download newDs2. Linux x86 binary, build 5.0-linux-20050422.

chmod +x newDs2, and copy into your bin directory.

Usage

newDs2 is used just like newDs, but typically you'll use the two additional options along with -includeSAM <wtsfile> to create a dataset containing only virtual channels. Note that it is very important to use the same bandwidth edges that you used with SAMcov to create the virtual channels, and you'll also want to use the -marker <name> and -time <start> <end> to specify the same segments of data you used with SAMcov. The result will contain one trial per event, properly filtered. It's a reasonable idea to widen the trial by a couple hundred milliseconds at the edges, to allow the filters to settle.

For example, if your covariance matrix was created using

SAMcov -m postvspre -r $ds -f "14 26"

where $ds/SAM/postvspre contains

1
stim 0 .5
1
stim -.5 0

(active post-, control pre-stim), and you created virtual channels using

SAMsrc -W 0 -t targets -r $ds -c postvspre,14-26Hz

then you can create a virtual channel dataset using

newDs2 -f -band 14 26 -excludeMEG -includeSAM postvspre,14-26Hz,targets.wts \
     -marker stim -time -.6 .6 $ds ${newname}.ds

Note that in this case, where the covariance is for a dual state, both active and control regions will be included in the covariance used for the virtual channel calculation (even if they are not adjacent). So, it is appropriate to create trials containing both regions (plus a little extra on the ends).

Note that you don't need to exclude the MEG channels, but the dataset will be smaller if you do.

Next you'll want to further process the virtual channel data by either averaging

averageDs ${newname}.ds ${newname}-av.ds

to observe evoked components, or computing an estimate of the power (using, e.g., hilbertDs or rmspowerDs) and then averaging, to observe induced components (see InducedEvoked). You can also use virtual channel datasets with ctf2st.