And even without an increase in SNR, on-scalp OPMs allow for higher spatial frequencies to be recorded and are expected to be more focal, which should benefit laminar inference.

I suppose at their current state of development, the SNR advantage of putting the sensors closer to the scalp will more or less be cancelled out by the differences in noise floors. But exciting to see where further innovations will take us!

My hunch, based on my simulations and @danclab.bsky.social's work demonstrating laminar inference in cryogenic MEG, is that current systems are already in the right ball park for the paradigms you mention. Highly accurate co-registration between OPM sensors and the MRI will be crucial, though!

While we are not there yet, it is encouraging that first experimental studies demonstrate that OPM-MEG systems, even at much lower sensor counts, show comparable signal amplitudes and source reconstruction accuracy to cryogenic MEG.

Nevertheless, as a guideline, it’s useful to think that for a given signal, we expect an ~ 5-fold increase in SNR over cryogenic MEG just by virtue of moving the sensors closer to the scalp—if we had the same number of sensors for both systems and comparable sensor noise floors (Boto et al. 2016).

Thanks for your interest, @hritz.bsky.social! I agree with @danclab.bsky.social, the SNR will depend a lot on your signal of interest, its spatial location and as you mention, on whether we look at single-trial or averaged data. And different OPM systems will show different sensitivity.

Thanks for the inspiration and all your support!

Hopefully we'll move from theoretically to experimentally soon!
And thanks so much for our discussions and your generous feedback!