rmspowerDs

RMS stands for root mean square. The rmspowerDs program will compute the RMS in a sliding window across your data. This is roughly the same as calculating the envelope of your data, and is important for estimating induced power levels over time (see InducedEvoked).

See also hilbertDs.

Important options you'll need:

-window n

This specifies the width of the sliding window in seconds. It needs to be short, a few cycles of your frequency of interest. For example, if you are looking at gamma band activity (30–50 Hz), one cycle is about 1/40 s or 25 ms. Thus a window width of about .075 may be appropriate. A larger window would lower the temporal resolution of your estimate, while a smaller window would be noisier.

-hanning

A Hanning window is small at the edges and big in the middle; it is multiplied by the data before the RMS is calculated. Thus, it emphasizes the center of the window, so you might want to widen the window slightly. Also, the result will look much smoother. The default is a square (i.e., no) window function.

The usual newDs options are also available. For example:

rmspowerDs -window .075 -hanning -marker stim -time -.5 .5 virt.ds virt-rms.ds

will create one 1 s trial per stim. This example assumes that virt.ds contains virtual channels that have already been filtered from 30–50 Hz. You can also use the -process, etc., options to filter on the fly.

Typically, the next step is to average the RMS across trials, to get the induced activity.

averageDs -marker stimA virt-rms.ds virt-stimA-induced.ds
averageDs -marker stimB virt-rms.ds virt-stimB-induced.ds

A comparison of the two averages will tend to reflect the results of a differential SAM analysis (-D3) for the same time window, but using rmspowerDs with virtual channels gives you better time resolution.