ENH: Implement method to plot DWI gradient

[sebastientourbier]

Providing the option to visualize the DWI gradient.

The diffusion gradient is displayed on a unit sphere where for each direction the size and color vary depending on its bvalue.

It works as the following:

The notebook can be found in docs/notebooks.

[pep8speaks]

Hello @sebastientourbier! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

• In the file eddymotion/viz.py:

Line 40:1: E303 too many blank lines (4)

Comment last updated at 2021-04-12 09:32:43 UTC

[arokem]

I think that the way to copy this code over is to include the original license somewhere in this file (you can also additionally include a note in every docstring pointing to the original implementation).

[oesteban]

I believe the BSD license does not mandate that (although, I agree what you propose would be the right thing to do).

I propose:

1. Retain the copyright notice at the top of the file Ariel liked, as mandated for redistribution of the original source code by BSD.
2. Make a list of the most relevant modifications of the code, as if this were some sort of Apache License.

[oesteban]

I'll send a proposal of how to do this as soon as I get the time.

[sebastientourbier]

Thanks @oesteban!

It is indeed not clear yet to me how to make it the most clear.

In summary:

• the method rotation_matrix is exactly the code of Emmanuel.
• the method draw_circle has been modified to take the full list of normalized bvecs and corresponding circle radii instead of taking the list of bvecs and radii for a specific shell (bvalue)
• the methods draw_points has been majorly modified including:
  • the input is a single 2D numpy array of the gradient table in RAS+B format
  • the scaling of the circle radius for each bvec proportional to the inverse of the bvals. A minimum/maximal value for the radii can be specified
  • Circles for each bvec are drawn at once instead of looping over the shells.
  • Some variables have been renamed (like vects to bvecs)

[oesteban]

A few suggestions from a first quick review.

[oesteban]

Suggested change

	figsize=(9.0, 9.0),
	ax=None,

[oesteban]

Suggested change

	fig = plt.figure(figsize=figsize)
	ax = fig.add_subplot(111, projection='3d')
	if ax is None:
	figsize = kwargs.pop("figsize", (9.0, 9.0))
	fig = plt.figure(figsize=figsize)
	ax = fig.add_subplot(111, projection='3d')

[oesteban]

@sebastientourbier I have sent a PR against this branch to address the licensing comments.

[oesteban]

Re-add some suggestions after my PR left them outdated

[oesteban]

Okay, I think after my latest suggestions we can go ahead and merge this. WDYT @sebastientourbier ?

[oesteban]

One more little thing I forgot I meant to suggest