Open ekellbuch opened 8 years ago
Since we don't know the ground truth of when spikes are arriving, we'll have to think of an unbiased way to calculate SNR. For instance, right now if we just ran the spike sorting module and calculated the power of all of the spikes detected vs the power of the signal when there is no spike, we probably would have roughly the same SNR for each of the channels since the threshold is adjusted on a channel by channel basis (4 * std deviation or something similar). We could probably do a test run first with the same threshold for each channel so we can compare SNRs across channels
estimate SNR and suggest channels to focus and/or avoid. This could speed up process for doing stimulation (e.g. single unit triggered stim), specially if high density MEA are available. It could also help with long term recordings.