FIX : fix ch_names in topomaps

[agramfort]

cc @mainakjas for #1548

[agramfort]

@mainakjas let me know if this works.

[coveralls]

Coverage increased (+0.0%) when pulling f1a1191 on agramfort:fix_names_topomap into e273498 on mne-tools:master.

[mainakjas]

Thanks @agramfort! hmm ... this seems to be fine for ch_type=mag but I still see missing channels on the periphery when ch_type=grad (e.g., MEG 0922).

[agramfort]

yes there is a bug as the grad names are not combined when computing their
norms.
What should be the name of the combined grads?

[mainakjas]

MEG 092x, MEG 161x etc for mne.viz.plot_evoked_topomap? I want to visualize some stats -- so combining norms may not be ideal for mne.viz.plot_topomap ...

[agramfort]

MEG 092x, MEG 161x etc for mne.viz.plot_evoked_topomap?

fine with me.

I want to visualize some stats -- so combining norms may not be ideal for mne.viz.plot_topomap ...

what's your suggestion? what is it exactly you want to visualize?

[mainakjas]

On Mon, Sep 8, 2014 at 5:54 PM, Alexandre Gramfort <[email protected]

wrote:

MEG 092x, MEG 161x etc for mne.viz.plot_evoked_topomap?

fine with me.

I want to visualize some stats -- so combining norms may not be ideal
for mne.viz.plot_topomap ...

what's your suggestion?

show_names = True should show all channels in mne.viz.plot_topomap() since
this allows arbitrary distribution of sensor positions. And show_names =
True should show MEG 092x, MEG 161x etc for the rest of the
plot_***_topomap() functions.

what is it exactly you want to visualize?

classification accuracies. I'd prefer to take just the mean (rather than
rms) or display both the grad channels. Currently, the easiest way to do
this appears to be EvokedArray -> plot_topomap.

—
Reply to this email directly or view it on GitHub
#1550 (comment).

[agramfort]

what's your suggestion?

show_names = True should show all channels in mne.viz.plot_topomap() since
this allows arbitrary distribution of sensor positions. And show_names =
True should show MEG 092x, MEG 161x etc for the rest of the
plot_***_topomap() functions.

the thing is that we automatically combine grads in the viz function.

what is it exactly you want to visualize?

classification accuracies. I'd prefer to take just the mean (rather than
rms) or display both the grad channels. Currently, the easiest way to do
this appears to be EvokedArray -> plot_topomap.

to me the cleanest thing would be to add a method to containers
(Epochs, Evoked) to combine grads and create a new container. This
method would take a param such as mse, mean, angle etc... We'd need a
new sensor type for combined sensors that is understood by topomap
functions...

[mainakjas]

On Mon, Sep 8, 2014 at 7:35 PM, Alexandre Gramfort <[email protected]

wrote:

what's your suggestion?

show_names = True should show all channels in mne.viz.plot_topomap()
since
this allows arbitrary distribution of sensor positions. And show_names =
True should show MEG 092x, MEG 161x etc for the rest of the
plot_***_topomap() functions.

the thing is that we automatically combine grads in the viz function.

what is it exactly you want to visualize?

classification accuracies. I'd prefer to take just the mean (rather than
rms) or display both the grad channels. Currently, the easiest way to do
this appears to be EvokedArray -> plot_topomap.

to me the cleanest thing would be to add a method to containers
(Epochs, Evoked) to combine grads and create a new container. This
method would take a param such as mse, mean, angle etc... We'd need a
new sensor type for combined sensors that is understood by topomap
functions...

sounds like a plan :) for now, I think I can live with mse.

—
Reply to this email directly or view it on GitHub
#1550 (comment).

[agramfort]

sounds like a plan :) for now, I think I can live with mse.

looking forward the moment you need it :)

[mainakjas]

On Tue, Sep 9, 2014 at 4:36 PM, Alexandre Gramfort <[email protected]

wrote:

sounds like a plan :) for now, I think I can live with mse.

looking forward the moment you need it :)

so should we merge this once the grad names are fixed? I'll make a note of
the combining grads method in a new issue.

—
Reply to this email directly or view it on GitHub
#1550 (comment).

[mainakjas]

oops, issue already exists: #1280 :)

[agramfort]

here we go. Done on my side.

[coveralls]

Coverage increased (+0.0%) when pulling 0730bbd on agramfort:fix_names_topomap into 3dbbb95 on mne-tools:master.

[coveralls]

Coverage increased (+0.0%) when pulling 0730bbd on agramfort:fix_names_topomap into 3dbbb95 on mne-tools:master.

[mainakjas]

works perfectly! thanks @agramfort