[ENH, MRG] Add intracranial electrode localization GUI #9586

alexrockhill · 2021-07-19T02:14:22Z

Add a GUI that allows users to find their intracranial electrode montage from a CT, an MRI and the Freesurfer recon-all from it, and an info object that supplies the names to be found and is return with a modified montage.

Closes #9559

Very rough version, no review yet.

larsoner · 2021-07-19T13:55:23Z

a raw object that supplies the names to be found and is return with a modified montage.

It should be slightly more general to make this an instance of info that gets modified in-place.

Very rough version, no review yet.

Let me know when to look and test! Ideally you would not need to tell me what to do and I can figure it out from the docs/tutorial etc.

alexrockhill · 2021-07-19T15:36:50Z

It should be slightly more general to make this an instance of info that gets modified in-place.

Yeah that's how it is in the code, the comments were a bit out-of-date

agramfort · 2021-07-22T06:50:32Z

mne/gui/tests/test_ieeg_locate_gui.py

+
+
+def test_ieeg_elec_locate_gui_display():
+	mne.gui.locate_ieeg(subject, subjects_dir=subjects_dir)


can you micmic clicks ? I suspect this test leads to limited coverage no?

Yes, hopefully that will be done soon, I've been sorting out the API (e.g. other PR) but this is next

alexrockhill · 2021-07-22T21:13:40Z

Still a work in progress, just for a demo, no need to review

alexrockhill · 2021-07-27T00:53:18Z

Here's a script for use on the misc data

import os.path as op
import numpy as np

import nibabel as nib
from dipy.align import resample

import mne

misc_path = mne.datasets.misc.data_path()
subject = 'sample_seeg'
subjects_dir = op.join(misc_path, 'seeg')

CT_orig = nib.load(op.join(misc_path, 'seeg', 'sample_seeg_CT.mgz'))
raw = mne.io.read_raw(op.join(misc_path, 'seeg', 'sample_seeg_ieeg.fif'))

T1 = nib.load(op.join(misc_path, 'seeg', 'sample_seeg', 'mri', 'T1.mgz'))
reg_affine = np.array([
    [0.99270756, -0.03243313, 0.11610254, -133.094156],
    [0.04374389, 0.99439665, -0.09623816, -97.58320673],
    [-0.11233068, 0.10061512, 0.98856381, -84.45551601],
    [0., 0., 0., 1.]])
CT_aligned = mne.transforms.apply_volume_registration(CT_orig, T1, reg_affine)

trans = mne.coreg.estimate_head_mri_t(subject, subjects_dir)
gui = mne.gui.locate_ieeg(raw.info, trans, CT_aligned, subject=subject, subjects_dir=subjects_dir)

alexrockhill · 2021-07-28T02:37:20Z

I didn't find anything that helpful in mne.viz.plot_bem but I figured out the transforms by following this https://github.com/moloney/dcmstack/blob/master/src/dcmstack/dcmstack.py#L246:L252

larsoner · 2021-07-28T12:18:39Z

I didn't find anything that helpful in mne.viz.plot_bem

It produces outputs like this:

https://mne.tools/stable/auto_tutorials/forward/30_forward.html#sphx-glr-auto-tutorials-forward-30-forward-py#compute-and-visualize-bem-surfaces

The slicing order and labeling of each axis has already been done in there somewhere. Following the trail:

mne-python/mne/viz/misc.py

Line 561 in 474da96

return _plot_mri_contours(mri_fname, surfaces, src, orientation, slices,
mne-python/mne/viz/misc.py

Lines 322 to 323 in 474da96

(x, y, z), (flip_x, flip_y, flip_z), order = _mri_orientation(

nim, orientation)

mne-python/mne/_freesurfer.py

Lines 84 to 97 in 474da96

    
           import nibabel as nib 
        
           _validate_type(img, nib.spatialimages.SpatialImage) 
        
           _check_option('orientation', orientation, ('coronal', 'axial', 'sagittal')) 
        
           axcodes = ''.join(nib.orientations.aff2axcodes(img.affine)) 
        
           flips = {o: (1 if o in axcodes else -1) for o in 'RAS'} 
        
           axcodes = axcodes.replace('L', 'R').replace('P', 'A').replace('I', 'S') 
        
           order = dict( 
        
               coronal=('R', 'S', 'A'), 
        
               axial=('R', 'A', 'S'), 
        
               sagittal=('A', 'S', 'R'), 
        
           )[orientation] 
        
           xyz = tuple(axcodes.index(c) for c in order) 
        
           flips = tuple(flips[c] for c in order) 
        
           return xyz, flips, order

And to see how data should be manipulated to plot properly, you want to look in _plot_mri_contours, though it gets a bit complicated:

mne-python/mne/viz/misc.py

Line 391 in 474da96

for ai, (ax, sl, lower, upper) in enumerate(zip(

Ideally you could just use _plot_mri_contours directly, for example, and improve it to have any additional functionality you need, probably via refactoring.

larsoner · 2021-07-28T12:24:33Z

... and note that doing it this way, the affine / vox2ras / vox2ras_tkr are those of the original volume, no messing around needed. You just plot such that you're in canonical orientation, and if someone clicks, you figure out which voxel that was in the volume (based on which slice you're plotting and whether or not it's flipped) and apply the affine/vox2ras/vox2ras_tkr from the original volume.

alexrockhill · 2021-07-28T17:35:45Z

mne-python/mne/viz/misc.py

Line 391 in 474da96

for ai, (ax, sl, lower, upper) in enumerate(zip(

Ideally you could just use _plot_mri_contours directly, for example, and improve it to have any additional functionality you need, probably via refactoring.

So I followed to that point to but what I really need is the affine transformation as you were saying. Those are all really convoluted ways of getting what nibabel.orientations.aff2axcodes and nibabel.orientations.axcodes2ornt gives you in one or two lines that should be refactored. I'm trying to focus on the GUI right now and refactoring plot_bem when it doesn't contribute to the GUI doesn't seem like a good use of time until after the main things are finished at which point I would be happy to go back.

This part is the key part that is kind of in plot_bem but it's spread over many lines of code in a very confusing way

ras_ornt = nib.orientations.axcodes2ornt('RAS')
ornt_trans = nib.orientations.ornt_transform(ornt, ras_ornt)
img_data = nib.orientations.apply_orientation(orig_data, ornt_trans)
orig_mgh = nib.MGHImage(orig_data, img.affine)
aff_trans = nib.orientations.inv_ornt_aff(ornt_trans, img.shape)
vox_ras_t = np.dot(orig_mgh.header.get_vox2ras_tkr(), aff_trans)

Again, I'd be happy to go back and refactor it in plot_bem but it's not going to help the GUI.

Just to make the point clear, flip_x which is an output of the _mri_orientation in _freesurfer.py is used four separate times. The code I need is the above and refactoring of plot_bem is going to take a lot of time and unrelated testing of the bem plotting without any crossover to the GUI.

alexrockhill · 2021-07-28T17:38:58Z

I'll try a separate PR to refactor the BEM code so that maybe it can share that code above maybe that will make things easier

larsoner · 2021-07-28T17:51:12Z

So I followed to that point to but what I really need is the affine transformation as you were saying... This part is the key part that is kind of in plot_bem but it's spread over many lines of code in a very confusing way

What I don't follow is what affine do you need? Why aren't you just using img.affine? What is your procedure for plotting data so that you need something else?

In the BEM code what we do is use the image's affine as is (if we need it), and then figure out how to slice into the volume to show standard orientations (sagittal, axial, coronal). To me this is a simple enough procedure and does what you should need for the GUI.

Perhaps if you describe conceptually what you do in the GUI and why you need some other affine it would help.

refactoring plot_bem when it doesn't contribute to the GUI doesn't seem like a good use of time until after the main things are finished at which point I would be happy to go back

The BEM code 1) resolves orientations so that they can be plotted sagittal/axial/coronal, 2) shows the slice number, 3) displays L/R/A/P/I/S to the sides, 4) has contours-from-surfaces built in, 5) has been unit-tested and user-tested to make sure that non-standard orientations are shown properly. These all seem like useful features for the GUI, no?

larsoner · 2021-07-28T18:01:22Z

... maybe you're talking about the mapping from GUI click->RAS? For this I would just figure out the click->volume index i,j,k, which is easy enough to get from knowing the sag/cor/ax slice that was clicked, where it was clicked, and the flip/slicing that was used in the current plot. Once you have this volume i,j,k, you just dot it with img.affine / img.header.get_vox2ras_tkr.

alexrockhill · 2021-07-28T18:03:02Z

What I don't follow is what affine do you need? Why aren't you just using img.affine? What is your procedure for plotting data so that you need something else?

I need an affine that goes from surface RAS to the RAS voxels of the reoriented image.

In the BEM code what we do is use the image's affine as is (if we need it), and then figure out how to slice into the volume to show standard orientations (sagittal, axial, coronal). To me this is a simple enough procedure and does what you should need for the GUI.

Perhaps if you describe conceptually what you do in the GUI and why you need some other affine it would help.

The BEM code 1) resolves orientations so that they can be plotted sagittal/axial/coronal, 2) shows the slice number, 3) displays L/R/A/P/I/S to the sides, 4) has contours-from-surfaces built in, 5) has been unit-tested and user-tested to make sure that non-standard orientations are shown properly. These all seem like useful features for the GUI, no?

Yes, that is what is needed for the GUI but the flips and coordinate frames are a mess and all of that functionality is already in the GUI.

I'll work on a PR to reorient the MRI surface first so that it's not reoriented four separate times.

larsoner · 2021-07-28T18:14:33Z

I need an affine that goes from surface RAS to the RAS voxels of the reoriented image.

Ahh I think this is the sticking point. In the BEM code no explicit reorientation is done. The image is just sliced (and maybe flipped) as necessary to show properly oriented views. To me this is simple enough, and easy (enough) to know which voxel i/j/k is clicked, then use the standard img.affine to get RAS.

But if there is some way to reorient that doesn't actually involve resampling the image (the current code doesn't need to) and makes things simpler, that's great! (And the BEM code should benefit from this -- DRY, one right way to do a given thing, etc.!)

alexrockhill · 2021-07-28T20:17:53Z

I think something is incompatible on Windows with the way I've implemented PyQt from the tests, it looks like there are segfaults in the tests. Any idea what's going on @larsoner or anyone else?

mne/gui/__init__.py

alexrockhill · 2021-08-10T20:48:57Z

Ok, I pushed this because I needed it for some analysis now that the coordinate frames are all worked out but it is almost feature complete.

Things still left to do:

Fix extra empty figure popping up from the renderer
Speed up initialization
Fix too many local maxima found (almost done)
Add tube for each automatically detected device with a sphere for which contact is currently selected
Add tests

I hope about a week away from being fully finished. We'll see though, some things can be quite finicky with the approach so we should test it on Liberty's data too and make sure everything works. In general it would be nice to avoid parameters about the contact size, spacing etc. as those will make it more confusing but they can limit the number of false positives so that takes some finagling to find a good balance.

mne/_freesurfer.py

agramfort · 2021-08-15T11:55:51Z

@alexrockhill do you need feedback from me here?

alexrockhill · 2021-08-15T21:20:36Z

@alexrockhill do you need feedback from me here?

Now is fine or I'm adding tests so I will change a few more things and maybe I will find my own errors first over the next day or two but it is very close to being done.

larsoner · 2021-09-08T15:03:16Z

upgrading pyvistaqt didn't seem to help ...

I would try in a fresh MNE environment created according to our current installation instructions. Also if you're on an ARM macOS machine then you probably need to use the i686 version of Anaconda rather than the native ARM one, I doubt all packages are available and working for ARM

jasmainak · 2021-09-13T03:43:01Z

Okay finally got this to work. Thanks @larsoner !

Disclaimer, I have never done this myself. So, this is mostly a naive user perspective.

Why is the CT/MRI image inside two rectangles?
I can't understand the bottom right plot. When I click on a marker, is it supposed to center and mark the corresponding channel in bottom right? That doesn't seem to be happening
Is there a way to turn off the overlay of the CT image over the MRI? Or does it only show CT?
I tried changing the color of one channel. Didn't seem to work ...
Can there be a button for help instead of just a keyboard shortcut? Ideally, folks should be able to run the GUI without reading too many instructions
Wasn't clear to me what "snap to center" does
If I try to "jump to a location" by editing the RAS coordinates, it does not work when I put a crazy location like 150. I guess it would be more intuitive if it had up/down arrows for each coordinate and these don't go past a certain number

alexrockhill · 2021-09-15T01:55:23Z

Okay finally got this to work. Thanks @larsoner !

Disclaimer, I have never done this myself. So, this is mostly a naive user perspective.

Why is the CT/MRI image inside two rectangles?

That's because the background is non-zero and the padding in matplotlib is zeros (black). It doesn't seem to distracting to me (you're zoomed in very close most of the time). Is it an issue to you?

I can't understand the bottom right plot. When I click on a marker, is it supposed to center and mark the corresponding channel in bottom right? That doesn't seem to be happening

The bottom-right plot is view only (clicking in 3D doesn't make sense, you are clicking based on what is in the other three plots, the 3D one is mostly for identifying devices). Clicking and scrolling on the plot only moves the camera.

Is there a way to turn off the overlay of the CT image over the MRI? Or does it only show CT?

The screenshot you sent is just of the CT. You can press 'b' to toggle the view of the brain. I should add back the help button though come to think of it.

I tried changing the color of one channel. Didn't seem to work ...

To change the "device" of a contact, you have to change the color and then mark the contact. I guess you could change the group of unmarked contacts as well.

Can there be a button for help instead of just a keyboard shortcut? Ideally, folks should be able to run the GUI without reading too many instructions

Yeah, I should add that back.

Wasn't clear to me what "snap to center" does

It snaps to the center of an area of high-intensity. The technical aspects are are that it looks for a local maxima and then includes all monotonically decreasing voxels from that voxel above 50% of the maximal value but from a typical user perspective the first part is all you really need to know.

If I try to "jump to a location" by editing the RAS coordinates, it does not work when I put a crazy location like 150. I guess it would be more intuitive if it had up/down arrows for each coordinate and these don't go past a certain number

If you put in something outside the range, it just resets to what it was before. That follows what Freeview does and I think it's not too hard to figure out with a few entries. It should be user-proof but I don't necessarily think you need warnings and errors for this when you can follow the existing format with small modifications to figure it out pretty easily.

jasmainak · 2021-09-15T04:24:05Z

The bottom-right plot is view only (clicking in 3D doesn't make sense, you are clicking based on what is in the other three plots, the 3D one is mostly for identifying devices). Clicking and scrolling on the plot only moves the camera.

Sorry, I should have said when I click on the marker name on the right panel, not on the plot. I thought it would somehow highlight the marker in the bottom right corner but it does nothing

alexrockhill · 2021-09-15T05:02:32Z

It is centered as the focal point in the view but for forward compatibility with an automated algorithm (which would also have a current selected contact) it doesn't have an indicator for the latest contact marked, you just have to go off what changed in the plot. This is because two markers would be confusing in the future when the current auto-detected contact needs to be indicated.

larsoner · 2021-09-15T18:30:37Z

That's because the background is non-zero and the padding in matplotlib is zeros (black). It doesn't seem to distracting to me (you're zoomed in very close most of the time). Is it an issue to you?

Can you systematically use/choose the border constant value as the median of the first/last i/j/k planes np.median(np.concatenate(img[[0, -1]].ravel(), img[:, [0, -1]].ravel(), img[:, :, [0, -1].ravel()) rather than just using zero? Seems like that would be robust and look nicer.

The bottom-right plot is view only (clicking in 3D doesn't make sense, you are clicking based on what is in the other three plots, the 3D one is mostly for identifying devices). Clicking and scrolling on the plot only moves the camera.

Maybe a title for this widget "3D View" would make it clearer that this is just a view and not something that can be interacted with via clicks.

The screenshot you sent is just of the CT. You can press 'b' to toggle the view of the brain. I should add back the help button though come to think of it.

In general in UI design, anything that can be done with a keyboard shortcut should have a corresponding mouse-based control (e.g., checkbox, radio button, drop-down, etc.) when possible/reasonable. In this case I think a "Brain" checkbox or something might be good.

larsoner · 2021-09-15T18:31:37Z

... also, aesthetically if the background of all the MRI images is black, the 3D image should also have a black background, unless it messes up the visualization of electrode colors.

alexrockhill · 2021-09-15T18:34:35Z

Can you systematically use/choose the border constant value as the median of the first/last i/j/k planes np.median(np.concatenate(img[[0, -1]].ravel(), img[:, [0, -1]].ravel(), img[:, :, [0, -1].ravel()) rather than just using zero? Seems like that would be robust and look nicer.

-1, I think this is solving a problem that doesn't exist and it's nice to know for me the boundaries of the CT so that you can see which parts got cutoff as sometimes they do.

Maybe a title for this widget "3D View" would make it clearer that this is just a view and not something that can be interacted with via clicks.

-1 I don't think that title really makes that clearer and I think directions for that would be better placed in the tutorial. I do want to hold the users hand, but if you don't read in the instructions, I don't think it has to have more instructions in the GUI, ideally there would be as little text as possible.

In general in UI design, anything that can be done with a keyboard shortcut should have a corresponding mouse-based control (e.g., checkbox, radio button, drop-down, etc.) when possible/reasonable. In this case I think a "Brain" checkbox or something might be good.

Sure that sounds helpful.

larsoner · 2021-09-15T18:38:07Z

-1, I think this is solving a problem that doesn't exist

Hmm... I don't like the way it looks now, and I think @jasmainak does not either, so I guess we both at least see it as an aesthetic issue.

it's nice to know for me the boundaries of the CT so that you can see which parts got cutoff as sometimes they do.

The way it's coded now, sometimes you'll have that information and other times you won't, depending on the normalization of the CT image. To me it's better to systematically fix the cval/border bug, then actually add a bounding box rectangle using a color of your choosing (red?) than rely on the cval to do this for you

larsoner · 2021-09-15T18:39:19Z

(But I think it's okay to fix this cval business in a follow-up PR, it's easy enough and should clearly show an isolated improvement / we can discuss once it's implemented if it's actually better)

alexrockhill · 2021-09-15T18:44:22Z

-1, I think this is solving a problem that doesn't exist

Hmm... I don't like the way it looks now, and I think @jasmainak does not either, so I guess we both at least see it as an aesthetic issue.

it's nice to know for me the boundaries of the CT so that you can see which parts got cutoff as sometimes they do.

The way it's coded now, sometimes you'll have that information and other times you won't, depending on the normalization of the CT image. To me it's better to systematically fix the cval/border bug, then actually add a bounding box rectangle using a color of your choosing (red?) than rely on the cval to do this for you

I'm fine with a bounding box, that seems like a reasonable solution, the point was just that where the image ends is important information. The CT isn't normalized though so it will usually be some non-zero low number.

getting closer to a working draft [skip ci] [skip circle] wip fixed some little things wip wip wip finally fixed the transform issue wip wip wip wip wip fix units wip working versions with separate plots fix coordinate frames, refactor surface.py change to this PR wip add 3d plot automatic detection wip almost feature complete a few fixes attempted fix but doesn't work on ubuntu for some reason [skip ci][skip circle] fix zoom fix draw bug allow no head lots of cleaning up fix, change to calling dipy directly feature complete version wip wip add snap feature wip fix tests fix tutorial fix flake fix thresh fix tests didn't save fix tests fix tests fix spelling fix test fix refs small bug fixes fix warnings clean up plotting, fix pernicious array overwriting bug wip wip C- version, finds 6 / 15 electrodes with most contacts fix bug, add ACPC alignment back to tutorial, outline Hough transform wip wip wip fix diff fix diff again wip remove auto-find for now fix tests wip review, add tests everything works great except compute_volume_registration for the MRI and CT, fixing that ASAP fix registration, remove link, fix test revert to older version Eric review increase cursor width fix tests fix subjects_dir fix tutorial fix tests fix tests Update mne/gui/_ieeg_locate_gui.py Update mne/gui/tests/test_ieeg_locate_gui.py FIX: pyvistaqt, not mayavi STY: Cleaner MAINT: pytest

alexrockhill · 2021-09-15T20:28:53Z

Ok I added buttons for help and show/hide brain. You still can only use keys to scroll slices but I'm not sure there's an easy place to put those six buttons but you can always click instead of scrolling. Also zoom is only scroll button or a keypress.

I didn't address the bounding box but that would be nice to do in a followup.

jasmainak · 2021-09-15T21:04:21Z

Sorry I got lost with the bounding box discussion but looking at the image it seemed to me that there was some affine registration issue but I might be wrong. I have seen those kinds of images when you do some registration and the corners become black because there are no pixels to copy from.

It might be worth taking inspiration from MMVT. Maybe you were also a contributor @alexrockhill ? ;-)

I like how in the sliced view the CT appears in a different color and makes it really easy to spot things. But I don't know if it's a standard way to do this ...

alexrockhill · 2021-09-15T21:12:32Z

Sorry I got lost with the bounding box discussion but looking at the image it seemed to me that there was some affine registration issue but I might be wrong. I have seen those kinds of images when you do some registration and the corners become black because there are no pixels to copy from.

It might be worth taking inspiration from MMVT. Maybe you were also a contributor @alexrockhill ? ;-)

I like how in the sliced view the CT appears in a different color and makes it really easy to spot things. But I don't know if it's a standard way to do this ...

This is how it was previously in img_pipe but there were issues with this method because of setting the threshold. It's better to see the un-thresholded image to see the extent of the contact as clearly as possible. This is because the intensities in the CT vary by location and how directly the pixel that is most directly on the contact aligns with the contact.

EDIT: for MMVT, Noam was looking mostly at DBS which is usually well-spaced so there aren't bridged contacts and they come out super clean in the thresholded, blue version but with the SEEG CT data I have been working with, the bridged contacts make it so that you really need the background and no thresholding to get the full picture of the contacts. From the experience of actually having done this process for an experiment, I can tell you that the method currently on this branch works better even if it isn't as aesthetically pleasing.

Looking more closely, the black rectangle immediately surrounding the skull and the gray rectangle surrounding that are actually part of the underlying image. I really would prefer not to mess with this and the background because you wouldn't want to obscure something that might help you understand what's going on in the CT and it's bounding. The outermost black background is what happens when you zoom out of a matplotlib image so that one could be fixed but again I would strongly prefer to leave it as is.

alexrockhill · 2021-09-15T22:27:37Z

Looks like the failure is unrelated. Maybe we can get this moving and make more improvements later

larsoner · 2021-09-15T23:00:51Z

Agreed, okay for merge and incremental PRs after this @jasmainak ?

alexrockhill · 2021-09-16T05:08:38Z

The plan is to do a JOSS paper after this merges so there will be a review there as well

agramfort

+1 for MRG if CIs are all green.

alexrockhill · 2021-09-17T00:17:10Z

I'm redoing Liberty's ECoG grids with it (I don't have any ECoG that I'm doing for my own research) and besides it not being automated, it works really well

jasmainak · 2021-09-17T05:15:56Z

yep +1 for MRG and incremental PRs. Thanks @alexrockhill for the massive effort !

alexrockhill · 2021-09-17T05:48:55Z

@larsoner did you mean to remove the directional labels (eg superior, inferior)? I didn't see them when I just used it.

larsoner · 2021-09-17T16:51:06Z

@larsoner did you mean to remove the directional labels (eg superior, inferior)? I didn't see them when I just used it.

No, I don't think I meant to do it -- sorry if I did so in some commit. Good for a next PR :)

larsoner · 2021-09-17T16:53:15Z

Thanks @alexrockhill !!!

agramfort reviewed Jul 22, 2021

View reviewed changes

alexrockhill force-pushed the gui branch from 73c25db to 9405705 Compare July 22, 2021 21:13

alexrockhill force-pushed the gui branch from 9405705 to b579c9c Compare July 27, 2021 00:44

alexrockhill force-pushed the gui branch from b579c9c to 9516e48 Compare July 27, 2021 05:05

alexrockhill force-pushed the gui branch from 9516e48 to cef387d Compare July 28, 2021 02:37

alexrockhill mentioned this pull request Jul 28, 2021

[MAINT, MRG] Refactor BEM code #9625

Merged

larsoner reviewed Jul 28, 2021

View reviewed changes

mne/gui/__init__.py Outdated Show resolved Hide resolved

This was referenced Aug 4, 2021

[BUG] trans estimation from mne._freesurfer.get_mni_fiducials differs from sample trans #9638

Closed

[ENH, MRG] Add mne.viz.Brain.plot_sensors and refactor mne.viz.plot_alignment #9585

Merged

alexrockhill force-pushed the gui branch from cef387d to ad559c7 Compare August 10, 2021 20:39

alexrockhill force-pushed the gui branch from ae98001 to f30d0cb Compare August 12, 2021 01:52

mmagnuski reviewed Aug 12, 2021

View reviewed changes

mne/_freesurfer.py Outdated Show resolved Hide resolved

alexrockhill force-pushed the gui branch from f30d0cb to ab1c357 Compare August 15, 2021 21:27

alexrockhill added 2 commits September 15, 2021 11:48

mainak + eric suggestions

57094f1

alexrockhill force-pushed the gui branch from 90455f0 to 57094f1 Compare September 15, 2021 20:26

agramfort approved these changes Sep 16, 2021

View reviewed changes

larsoner merged commit 4375989 into mne-tools:main Sep 17, 2021

larsoner mentioned this pull request Sep 17, 2021

[ENH] Add electrode coregistration GUI for aligned CT/MRIs for ieeg processing #9155

Closed

alexrockhill deleted the gui branch September 17, 2021 20:36

TomDonoghue mentioned this pull request Dec 11, 2021

[REVIEW]: Intracranial Electrode Location and Analysis in MNE-Python openjournals/joss-reviews#3897

Closed

40 tasks



		def test_ieeg_elec_locate_gui_display():
		mne.gui.locate_ieeg(subject, subjects_dir=subjects_dir)

[ENH, MRG] Add intracranial electrode localization GUI #9586

[ENH, MRG] Add intracranial electrode localization GUI #9586

Conversation

alexrockhill commented Jul 19, 2021 • edited by larsoner Loading

larsoner commented Jul 19, 2021

alexrockhill commented Jul 19, 2021

agramfort Jul 22, 2021

Choose a reason for hiding this comment

alexrockhill Jul 22, 2021

Choose a reason for hiding this comment

alexrockhill commented Jul 22, 2021

alexrockhill commented Jul 27, 2021 • edited Loading

alexrockhill commented Jul 28, 2021

larsoner commented Jul 28, 2021 • edited Loading

larsoner commented Jul 28, 2021

alexrockhill commented Jul 28, 2021 • edited Loading

alexrockhill commented Jul 28, 2021

larsoner commented Jul 28, 2021

larsoner commented Jul 28, 2021

alexrockhill commented Jul 28, 2021

larsoner commented Jul 28, 2021

alexrockhill commented Jul 28, 2021

alexrockhill commented Aug 10, 2021

agramfort commented Aug 15, 2021

alexrockhill commented Aug 15, 2021

larsoner commented Sep 8, 2021

jasmainak commented Sep 13, 2021 • edited Loading

alexrockhill commented Sep 15, 2021 • edited Loading

jasmainak commented Sep 15, 2021

alexrockhill commented Sep 15, 2021

larsoner commented Sep 15, 2021

larsoner commented Sep 15, 2021

alexrockhill commented Sep 15, 2021

larsoner commented Sep 15, 2021

larsoner commented Sep 15, 2021

alexrockhill commented Sep 15, 2021

alexrockhill commented Sep 15, 2021

jasmainak commented Sep 15, 2021

alexrockhill commented Sep 15, 2021 • edited Loading

alexrockhill commented Sep 15, 2021

larsoner commented Sep 15, 2021

alexrockhill commented Sep 16, 2021

agramfort left a comment

Choose a reason for hiding this comment

alexrockhill commented Sep 17, 2021

jasmainak commented Sep 17, 2021

alexrockhill commented Sep 17, 2021

larsoner commented Sep 17, 2021

larsoner commented Sep 17, 2021

alexrockhill commented Jul 19, 2021 •

edited by larsoner

Loading

alexrockhill commented Jul 27, 2021 •

edited

Loading

larsoner commented Jul 28, 2021 •

edited

Loading

alexrockhill commented Jul 28, 2021 •

edited

Loading

jasmainak commented Sep 13, 2021 •

edited

Loading

alexrockhill commented Sep 15, 2021 •

edited

Loading

alexrockhill commented Sep 15, 2021 •

edited

Loading