spatio-temporal cluster on sensor level?

✉️ Mailing List Archive (read-only)
1

Hello MNE (python) user,

I have another question related to mne_python v.08

I was wondering if there is a script for running Permutation t-test with
spatio-temporal clustering on sensor data?

I found one for source level (which looks brilliant), however for sensor
level, I found "only" temporal clustering on 1 specific sensor.

Thanks for your help

Elisabeth

2

Dear Elisabeth,

with regard to data organization, you can proceed exactly as with the
source space examples, just replace vertices with channels.
The critical difference amounts to how to define the neighboring spatiall
features, a.k.a. connectivity.
To compute this information you should use the FieldTrip neigbour
definitions. You can read those using this function:
http://martinos.org/mne/stable/generated/mne.read_ch_connectivity.html#mne.read_ch_connectivity

for exmple like this for a 4D/BTi 248 channels magnetometer system:

neighbor_file = 'fieldtrip/template/neighbours/bti248_neighb.mat'

connectivity = mne.channels.read_ch_connectivity(neighbor_file)

In case you have missing channels, you can define those to return the
connectivity matrix of the subset of remaining channels:

connectivity = mne.channels.read_ch_connectivity(neighbor_file, picks=picks)

The rest is the same, make sure your data matrices match the contrast type
(F vs T test) and make sure that the last dimension carries the spatial
information.

I hope that helps (we'll soon add a corresponding example).

Best,
Denis

3

hi,

it would be great to have an example for this feature. Any volunteer?

Cheers,
Alex