[MRG] Glass brain visualisation #245

lesteve · 2014-09-26T09:57:48Z

This pull request is WIP for now and was only opened to get some feedback, both visualisation-wise and code-wise.

Here is how it looks like atm:

The code to generate the plots is there.

GaelVaroquaux · 2014-09-26T10:40:56Z

Exciting! Cc @chrisfilo

agramfort · 2014-09-26T11:05:14Z

indeed ! really nice !

chrisgorgo · 2014-09-27T02:11:13Z

Indeed very exciting! Where is the outline coming from?

lesteve · 2014-09-27T06:48:09Z

The outline comes from an SVG with a little bit of work to extract the coordinates of the nodes from the SVG and draw the Bézier curves in matplotlib.

The SVG was created by Gaël by vectorising and simplifying a bitmap picture.

AlexandreAbraham · 2014-09-29T19:26:42Z

This is very nice indeed. Little remark: on the second and third images, something is visible outside of the brain, maybe we should crop that automatically?

Have you shown these images to neuroscientists? They usually have very specific needs and, from my experience, they ask for a high threshold so that they can see the anatomical features in the background. Are the guidelines overlayed on the brain sufficient for them?

lesteve · 2014-09-29T20:52:46Z

I was wondering about where these data points outside the brain were coming from and I was told it may well be the eyes. Those regions can be seen in the plotting examples too for example this one:
.

If there is a way to remove these regions automatically, it's probably worth doing. Something interesting to note, is that the values are pretty high (in this case very negative) and setting a high threshold won't make them go away.

Definitely a good idea to get some feedback from neuroscientists at one point in the future.

GaelVaroquaux · 2014-09-29T20:55:57Z

If there is a way to remove these regions automatically, it's probably worth doing.

I am against removing them: we are doing data viz, we shouldn't be
hidding features of the data, but showing it.

lesteve · 2014-09-29T20:59:33Z

That's a valid point indeed but then maybe it's still worth removing these regions in an ad-hoc way in our examples.

bthirion · 2014-09-29T21:13:12Z

+1 for keeping them. Maybe the subject closed the eyes while pressing the button.

bthirion · 2014-09-29T21:19:38Z

nilearn/plotting/glass_brain_scripts/plot_glass_brain.py

+data[np.isnan(data)] = 0
+
+bg_img, black_bg, vmin, vmax = img_plotting._load_anat()
+slicer = img_plotting.plot_glass_brain(bg_img, vmin=3000, vmax=2*310983,


What do these values come from ? It's a bit disturbing for users who don't understand them.

I tweaked these values by hand to have a nicer color range. To be honest, the only use of this plot is to align the brain schematics with the anatomy and eventually hand-tweak the SVG to make them fit a bit better. I don't think it will stay as an example plot.

chrisgorgo · 2014-09-30T16:47:53Z

@margulies how does this outline looks from a neuroanatomical point of view?

bthirion · 2014-09-30T16:54:37Z

nilearn/plotting/img_plotting.py

@@ -572,4 +573,77 @@ def plot_stat_map(stat_map_img, bg_img=MNI152TEMPLATE, cut_coords=None,
                               **kwargs)
    return slicer

+def plot_glass_brain(stat_map_img, #bg_img=MNI152TEMPLATE, cut_coords=None,
+                     output_file=None, #display_mode='ortho',TODO support display_mode


pending TODO

There are definitely a few of those TODOs sprinkled in the code. I'll add a note to myself to make sure I tackled all of them before merging.

lesteve · 2014-10-07T09:10:20Z

nilearn/plotting/slicers.py

@@ -832,11 +890,73 @@ class YZSlicer(OrthoSlicer):
               z=ZSlicer)


+class OrthoProjector(OrthoSlicer):


I am still inheriting from the cut-based slicer and disabling a few methods which is rather hacky. There must be a better way to structure this code, suggestions welcome!

I am still inheriting from the cut-based slicer and disabling a few methods
which is rather hacky. There must be a better way to structure this code,
suggestions welcome!

I would say: add a base class, and inherit from it in the OrthoSlicer,
where you add methods, and in your own projector, where you add different
methods.

GaelVaroquaux · 2014-10-07T12:14:46Z

You should paste here the results of the latest iteration of the code.

GaelVaroquaux · 2014-10-07T12:16:23Z

nilearn/plotting/glass_brain/brain_schematics.py

+    def _codes_segment(self, pts):
+        return [Path.MOVETO, Path.LINETO]
+
+    @staticmethod


Is there a reason to have a staticmethod, rather than a function, here? (ie something not linked to the class).

The main reason is that this is a helper function which is only used within the class. I could change it to a module-level function but you would then lose the hint.

The main reason is that this is a helper function which is only used
within the class. I could change it to a module-level function but you
would then lose the hint.

But it's not a method. An object is more than a namespace of methods. It
has an internal state and methods act on its internal state. That's why
reading code that is object-oriented is harder to read: you never know
what has changed and what has not.

Thus, I would really push for using functions rather than staticmethods.

Yeah I think we quickly chatted about that already before. I just see this low-level helper functions as not important enough to warrant polluting the module namespace. During a first read of the code, I feel it is easier to ignore these unnecessary details if they are nested within the class they are related to.

lesteve · 2014-10-07T13:36:07Z

Same plots with an alpha of 0.7:

lesteve · 2014-10-10T08:10:52Z

@GaelVaroquaux I think I tackled the comments you had the other day

GaelVaroquaux · 2014-10-10T08:29:14Z

nilearn/plotting/slicers.py

@@ -6,8 +6,10 @@
 """

 import operator
+import os


Unused import

You should use pyflakes in your editor, it is great to detect these things. See http://scipy-lectures.github.io/advanced/debugging/index.html#a-type-as-go-spell-checker-like-integration

Actually something like this configured already. I just didn't spot this one.

GaelVaroquaux · 2014-10-10T08:30:06Z

We should rename 'slicers.py'. But we need a better name. Any suggestions?

GaelVaroquaux · 2014-10-10T08:31:26Z

nilearn/plotting/slicers.py


-    def draw_cut(self, cut, data_bounds, bounding_box,
-                  type='imshow', **kwargs):
+    def draw(self, cut, data_bounds, bounding_box,


Good idea to rename the method!

I think that I would prefer it to be named something like 'draw_volume'. Also, we should probably make it private: it's for the internals of the object, not for end-users to call.

Please disregard the comment on making it private: I hadn't seen that it was in our Axes object. End users are not meant to use this object anyhow, so let's not bother.

The thing is that draw actually draws some 2d data. I could rename it draw_2d to make it more explicit.

As a side comment I also plan to go over the file and remove the cut* occurrences where easily doable.

The thing is that draw actually draws some 2d data. I could rename it draw_2d
to make it more explicit.

Yes, that's a good suggestion.

As a side comment I also plan to go over the file and remove the cut*
occurrences where easily doable.

Good idea. Thanks!

lesteve · 2014-10-10T08:34:40Z

We should rename 'slicers.py'. But we need a better name. Any suggestions?

I was thinking about displays.py

* Use it in the glass brain plot by default * Minort tweaks in plot_glass_brain.py to use the default color map

Otherwise the colormap goes gray towards the low alpha values

depending on black_bg. Also black_bg default to False because 'auto' makes sense only when plotting an anat. Updated plot_glass_brain.py accordingly.

- Only three kind of stroke-widths: 1.2, 2.5, 3 - Did some ungrouping in brain_schematics_top.svg - Regenerated JSONs

in brain_schematics_top.svg. Regenerated JSON

and regenerated JSONs

* transform_2d -> transform_to_2d * axes_class -> _axes_class * other minor stuff

Also: * create_display_fun renamed to display_factory * removed unused import * PEP8 fix

Also other minor tweaks

in plot_glass_brain in order to be aligned with brain schematics

This is the value that was used in all the latest screen snapshots from the PR

This script was mostly useful to produce the snapshots in the PR

GaelVaroquaux · 2014-12-04T13:40:45Z

I saw that, but I prefer it with the alpha.

lesteve · 2014-12-04T15:21:38Z

OK merging this long standing one !

[MRG] Glass brain visualisation

GaelVaroquaux · 2014-12-04T16:37:45Z

Fucking Ay!

Can you rebuild the online docs, please;

added output_image argument to plot_design_matrix

bthirion reviewed Sep 29, 2014
View reviewed changes

bthirion reviewed Sep 30, 2014
View reviewed changes

lesteve added Plotting The issue is related to plotting functionalities. New feature labels Oct 1, 2014

lesteve reviewed Oct 7, 2014
View reviewed changes

GaelVaroquaux reviewed Oct 7, 2014
View reviewed changes

lesteve force-pushed the glass-brain branch 3 times, most recently from d772b01 to 080b920 Compare October 8, 2014 13:31

lesteve force-pushed the glass-brain branch from aeb11d5 to 9909e21 Compare October 10, 2014 08:22

GaelVaroquaux reviewed Oct 10, 2014
View reviewed changes

lesteve added 22 commits December 4, 2014 14:20

Add transparent color maps using the full alpha range

5aa0a6a

* Use it in the glass brain plot by default * Minort tweaks in plot_glass_brain.py to use the default color map

Removal of redundant if statement

31ce2c9

The color should stay constant for transparent colormap

6cada3a

Otherwise the colormap goes gray towards the low alpha values

Use t-maps rather than contrasts in plot_glass_brain.py

7d17f98

Use max(abs(data)) for 2d projection rather than max(data)

d4d51e4

Change default plot_glass_brain colormap

f5ef98c

depending on black_bg. Also black_bg default to False because 'auto' makes sense only when plotting an anat. Updated plot_glass_brain.py accordingly.

Simplified glass brain SVGs

66dfaa4

- Only three kind of stroke-widths: 1.2, 2.5, 3 - Did some ungrouping in brain_schematics_top.svg - Regenerated JSONs

Remove invisible path

9f011d7

in brain_schematics_top.svg. Regenerated JSON

Thiner lines in the glass brain SVGs

4e1a37b

and regenerated JSONs

Use auto threshold for glass brain

927fa06

Tweak glass brain example in plot_demo_plotting.py

c2a6bf9

Add inferior temporal sulcus

7bcf1a2

Rename some lingering slicer to display

8a42985

Fix max instead of min typo in auto vmin logic

9c8f0b6

Cosmits

f37072b

* transform_2d -> transform_to_2d * axes_class -> _axes_class * other minor stuff

Add module docstrings

3e25fe6

Document more parameters in _plot_img_with_bg

d971917

Also: * create_display_fun renamed to display_factory * removed unused import * PEP8 fix

Added OrthoProjector smoke test

71c8111

Add anatomy in plot_align_svg.py

4db0995

Also other minor tweaks

Explicitly say that input data needs to be in MNI space

bc75c30

in plot_glass_brain in order to be aligned with brain schematics

Use alpha=0.7 by default for the brain schematics

a58cc15

This is the value that was used in all the latest screen snapshots from the PR

Remove plot_glass_brain.py

88f7036

This script was mostly useful to produce the snapshots in the PR

lesteve force-pushed the glass-brain branch from 3380114 to 88f7036 Compare December 4, 2014 13:30

lesteve added a commit that referenced this pull request Dec 4, 2014

Merge pull request #245 from lesteve/glass-brain

d947db4

[MRG] Glass brain visualisation

lesteve merged commit d947db4 into nilearn:master Dec 4, 2014

lesteve deleted the glass-brain branch December 5, 2014 12:56

kchawla-pi pushed a commit to kchawla-pi/nilearn that referenced this pull request Jan 2, 2020

Merge pull request nilearn#245 from bthirion/plot_design_matrix_output

bb3e9b1

added output_image argument to plot_design_matrix

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[MRG] Glass brain visualisation #245

[MRG] Glass brain visualisation #245

lesteve commented Sep 26, 2014

GaelVaroquaux commented Sep 26, 2014

agramfort commented Sep 26, 2014

chrisgorgo commented Sep 27, 2014

lesteve commented Sep 27, 2014

AlexandreAbraham commented Sep 29, 2014

lesteve commented Sep 29, 2014

GaelVaroquaux commented Sep 29, 2014

lesteve commented Sep 29, 2014

bthirion commented Sep 29, 2014

bthirion Sep 29, 2014

lesteve Sep 30, 2014

chrisgorgo commented Sep 30, 2014

bthirion Sep 30, 2014

lesteve Sep 30, 2014

lesteve Oct 7, 2014

GaelVaroquaux Oct 7, 2014

GaelVaroquaux commented Oct 7, 2014

GaelVaroquaux Oct 7, 2014

lesteve Oct 7, 2014

GaelVaroquaux Oct 7, 2014

lesteve Oct 7, 2014

lesteve commented Oct 7, 2014

lesteve commented Oct 10, 2014

GaelVaroquaux Oct 10, 2014

lesteve Oct 10, 2014

GaelVaroquaux commented Oct 10, 2014

GaelVaroquaux Oct 10, 2014

GaelVaroquaux Oct 10, 2014

lesteve Oct 10, 2014

GaelVaroquaux Oct 10, 2014

lesteve commented Oct 10, 2014

GaelVaroquaux commented Dec 4, 2014

lesteve commented Dec 4, 2014

GaelVaroquaux commented Dec 4, 2014

		@@ -832,11 +890,73 @@ class YZSlicer(OrthoSlicer):
		z=ZSlicer)


		class OrthoProjector(OrthoSlicer):

[MRG] Glass brain visualisation #245

[MRG] Glass brain visualisation #245

Conversation

lesteve commented Sep 26, 2014

GaelVaroquaux commented Sep 26, 2014

agramfort commented Sep 26, 2014

chrisgorgo commented Sep 27, 2014

lesteve commented Sep 27, 2014

AlexandreAbraham commented Sep 29, 2014

lesteve commented Sep 29, 2014

GaelVaroquaux commented Sep 29, 2014

lesteve commented Sep 29, 2014

bthirion commented Sep 29, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

chrisgorgo commented Sep 30, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

GaelVaroquaux commented Oct 7, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

lesteve commented Oct 7, 2014

lesteve commented Oct 10, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

GaelVaroquaux commented Oct 10, 2014

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

lesteve commented Oct 10, 2014

GaelVaroquaux commented Dec 4, 2014

lesteve commented Dec 4, 2014

GaelVaroquaux commented Dec 4, 2014