search

martindinov / hackathon

Where we will store all our hackathon-related code and files.
0 stars 2 forks source link

EEG processing and triggers for action #7

Open nmdanr2b opened 9 years ago

nmdanr2b commented 9 years ago

a) basic artifacting (removing obvious high amplitude spikes due to muscle movements/eye blinks) b) Setting custom band limits (could be just comparing "low frequency" and "high frequency", hence two bands only) c) custom set thresholds per band to serve as triggers for neurofeedback and neurostimulation (if band X stays above threshold Y micro volts for Z seconds then do <action - light a LED, play pre-defined sound, display picture if some monitor is connected, apply neurostimulation protocol of choice)
d) marking in the saved data and/or a separate log file when a threshold has been exceeded or the band amplitude went below the threshold after being above it etc. Overall, there should be appropriate marking in the saved data and/or separate log file every time a trigger has been activated and triggered action took place and then finished. e) comparison of band power between bands as a trigger for neurostimulation/feedback (for example, if alpha/theta ratio exceeds 1:2 then do X) f) comparison of band power between sensors as a trigger for neurostimulation/feedback (for example, if alpha power measured at sensor A in a position B exceeds alpha power at sensor D in a position C then do X)

Theoretically, if / then action trigger statements can be as complex as we want being derived from c), e), f) or other (for example coherence-decoherence related) selected thresholds simultaneously - the capability is inherently there.

For proof-of-concept at hackaton, implementing just c) will be sufficient, but is essential - see the "Sound and light" issue

nmdanr2b commented 9 years ago

In the future, better artifacting and ERP analysis are to be considered.

nmdanr2b commented 9 years ago

For c) it could be both "for Z seconds continuously", or "for Z seconds per predefined period of time" (for example, 1 minute). The same applies to e) and f)

Note that apart from being extremely useful for neurofeedback and potentially evoked potentials studies, the triggers outlined are fantastic for detecting and analysing neurostimulation effects (as in light a specific LED or play sound if X happens, clearly indicating the effect and leaving a mark in a log file/saved data).

nmdanr2b commented 9 years ago

It will be also handy to implement two triggers per band or, more generally speaking, per event class - "lower than" and "higher than". For example, "if an amplitude of a selected band is lower than X turn on a blue LED, if it is higher than Y - turn on red LED".