Mainak Jas comments

Results 645 comments of


                                            Mainak Jas

Question on workflow regarding continuous time scalp EEG

1. It should be determined by `ar.consensus_` that basically says what fraction of channels need to be bad for the epoch to be considered bad. You can just do `Autoreject(n_interpolates=0,...

multi channel support for ransac

okgithub: matching issues with contributors :)

multi channel support for ransac

hehe ;)

Add bad sensors count per epochs to reject log

can you add a test?

Detection of "Spikes" in sEEG Data

Hi Nebras -- autoreject is unsupervised, so there are no guarantees of course. I am guessing that it will remove only larger artifacts such as movement and blinks. Note that...

deal with flat channels

yes detection is easier for flat channels. But you still can't repair locally flat sensors using epoch-wise interpolation in MNE.

get_rejection_threshold stops optimizing threshold as soon as one threshold leads to rejection of all channels in at least one fold

humm interesting, that sounds like a bug to me. I think I verified before merging the [pull request to speed up using bayesian optimization](https://github.com/autoreject/autoreject/pull/77) that it gave the same result...

get_rejection_threshold stops optimizing threshold as soon as one threshold leads to rejection of all channels in at least one fold

ummm ... I think the simplest might be to leave out X which give `-Inf` for any fold since they correspond to a very low threshold (= rejecting too many...

get_rejection_threshold stops optimizing threshold as soon as one threshold leads to rejection of all channels in at least one fold

@thht can you share a dataset on which we can reproduce this?

get_rejection_threshold stops optimizing threshold as soon as one threshold leads to rejection of all channels in at least one fold

okay! no rush and thank you for this :)