vferat/pycrostates

Cross-validation / collaboration

Opened this issue · 2 comments

DominiqueMakowski commented

Hey folks, I stumbled on this repo thanks to @mscheltienne, and it looks great! I'm happy to see more work being done on microstates :)

I simply wanted to point out that we did a bit of work on that a few months ago for neurokit, comparing the different existing packages and going back to the original formulations and equations.

https://github.com/neuropsychology/NeuroKit/tree/master/neurokit2/microstates

See example here

As I see you're actively working on this, I'd be very interested if you guys could eventually run some validation or benchmarking so that we can compare our implementations, to make sure they are correct and the most efficient. It's always good to double-check

Let me know what you think!

mscheltienne commented

Hey, sorry for the late reply, and I'll let Victor comment if he disagrees, but I think it would be great to compare and cross-validate the 2 implementations. However, we are a bit short on time for now, and personally, I won't have time to allocate to this for now.

I think a good first step would be to pick a common dataset (MNE dataset?) and write the clustering code for it in both neurokit2 and pycrostates.

vferat commented

Hey Dominique,

Thanks for reaching us ! Indeed It would be a good idea to collaborate on this.
I recently implemented a fetcher for the mind-brain-body dataset with more than 200 pre-processed EEG available. We could use it to run some validation and benchmarking.

If you are interested, you can reach me by email at victor.ferat@unige.ch so we can organize a virtual meeting to discuss this in more details.

  • 👍1