I am a relative beginner with MNE (and EEG) and I don't know what's the
recommended way to deal with continuous data. I'm using a sort of
homebuilt 2-channel system, and I have whole data "sessions" (a few
minutes long for each file) where I am trying to plot the alpha over
time. There are no epochs or events to speak of, so I'm not sure if I
should just mark each session as a whole epoch, and then run time-freq
analysis on each one separately? So far, I just import my data, create a
Raw object+info, then filter for alpha. But I'd like to plot alpha over
time in a relatively smooth manner, to be able to compare "sessions",
not cut my data up into events. Any suggestions are appreciated! Thank
you!
It sounds like trying out `EpochsArray` might be a solution. You'd need to
make sure each session is the same length but it'd let you browse them
relatively easily. Another option would be to treat the multiple sessions
as different channels (even though they're the same channels being
recorded). This way you could continuously display multiple sessions at
once (e.g., you could call your channels `channel_1_session_1` and
`channel_1_session_2` to plot them side by side)
I would favor Chris' first (conventional), but not second (non
conventional) idea (sorry Chris;)).
Additionally, you can directly apply time_frequency decomposition on
raw.get_data(), using the *_array functions which work on numpy arrays.
Finally, if you're on the dev branch, (or willing to wait until the next
releast, which should happens in the next couple of days/weeks), you can
get some inspiration from our new example where we decode continuously
varying data from oscillatory activity: https://martinos.org/mne/dev/auto_examples/decoding/plot_decoding_spoc_CMC.html
There, we filter the raw signals, and use `make_fixed_length_events` to
segment the continuous raw signal into many small epochs that are easy to
analyze and plot