7. Plotting brain images

In this section, we detail the general tools to visualize neuroimaging volumes and surfaces with nilearn.

Nilearn comes with plotting function to display brain maps coming from Nifti-like images, in the nilearn.plotting module.

7.1. Different plotting functions

Nilearn has a set of plotting functions to plot brain volumes that are fined tuned to specific applications. Amongst other things, they use different heuristics to find cutting coordinates.

plot_anat

plot_anat
Plotting an anatomical image

plot_epi

plot_epi
Plotting an EPI, or T2* image

plot_glass_brain

plot_glass_brain
Glass brain visualization. By default plots maximum intensity projection of the absolute values. To plot positive and negative values set plot_abs parameter to False.

plot_stat_map

plot_stat_map
Plotting a statistical map, like a T-map, a Z-map, or an ICA, with an optional background

plot_roi

plot_roi
Plotting ROIs, or a mask, with an optional background

plot_connectome

plot_connectome
Plotting a connectome

Functions for automatic extraction of coords based on brain parcellations useful for plot_connectome are demonstrated in Example: Comparing connectomes on different reference atlases

plot_nodes

plot_markers
Plotting network nodes (markers)

Function for automatic plotting of network nodes (markers) and color coding them according to provided nodal measure (i.e. connection strength) as demonstrated in Example: Comparing connectomes on different reference atlases

plot_prob_atlas

plot_prob_atlas
Plotting 4D probabilistic atlas maps

plot_carpet

plot_carpet
Plotting voxel intensities across time.

plot_img

plot_img
General-purpose function, with no specific presets

Warning

Opening too many figures without closing

Each call to a plotting function creates a new figure by default. When used in non-interactive settings, such as a script or a program, these are not displayed, but still accumulate and eventually lead to slowing the execution and running out of memory.

To avoid this, you must close the plot as follow:

from nilearn import plotting
display = plotting.plot_stat_map(img)
display.close()

7.2. Different display modes

plot_ortho

display_mode="ortho", cut_coords=[36, -27, 60]
Ortho slicer: 3 cuts along the x, y, z directions

plot_z_many

display_mode="z", cut_coords=5
Cutting in the z direction, specifying the number of cuts

plot_x

display_mode="x", cut_coords=[-36, 36]
Cutting in the x direction, specifying the exact cuts

plot_y_small

display_mode="y", cut_coords=1
Cutting in the y direction, with only 1 cut, that is automatically positioned

plot_z_small

display_mode="z", cut_coords=1, colorbar=False
Cutting in the z direction, with only 1 cut, that is automatically positioned

plot_xz

display_mode="xz", cut_coords=[36, 60]
Cutting in the x and z direction, with cuts manually positioned

plot_yx

display_mode="yx", cut_coords=[-27, 36]
Cutting in the y and x direction, with cuts manually positioned

plot_yz

display_mode="yz", cut_coords=[-27, 60]
Cutting in the y and z direction, with cuts manually positioned

plot_tiled

display_mode="tiled", cut_coords=[36, -27, 60]
Tiled slicer: 3 cuts along the x, y, z directions, arranged in a 2x2 grid

plot_mosaic

display_mode="mosaic"
Mosaic slicer: multiple cuts along the x, y, z directions, with cuts automatically positioned by default

plot_lzr

Glass brain display_mode="lzr"
Glass brain and Connectome provide additional display modes due to the possibility of doing hemispheric projections. Check out: 'l', 'r', 'lr', 'lzr', 'lyr', ‘lzry', 'lyrz'.

plot_lyrz

Glass brain display_mode="lyrz"
Glass brain and Connectome provide additional display modes due to the possibility of doing hemispheric projections. Check out: 'l', 'r', 'lr', 'lzr', 'lyr', ‘lzry', 'lyrz'.

7.3. Available Colormaps

Nilearn plotting library ships with a set of extra colormaps, as seen in the image below

../_images/sphx_glr_plot_colormaps_001.png

These colormaps can be used as any other matplotlib colormap.

../_images/sphx_glr_plot_colormaps_002.png

7.4. Adding overlays, edges, contours, contour fillings, markers, scale bar

To add overlays, contours, or edges, use the return value of the plotting functions. Indeed, these return a display object, such as the nilearn.plotting.displays.OrthoSlicer. This object represents the plot, and has methods to add overlays, contours or edge maps:

display = plotting.plot_epi(...)

plot_edges

display.add_edges(img)
Add a plot of the edges of img, where edges are extracted using a Canny edge-detection routine. This is typically useful to check registration. Note that img should have some visible sharp edges. Typically an EPI img does not, but a T1 does.

plot_contours

display.add_contours(img, levels=[0.5], colors="r")
Add a plot of the contours of img, where contours are computed for constant values, specified in ‘levels’. This is typically useful to outline a mask, or ROI on top of another map.
Example: Plot Haxby masks

plot_fill

display.add_contours(img, filled=True, alpha=0.7, levels=[0.5], colors="b")
Add a plot of img with contours filled with colors

plot_overlay

display.add_overlay(img, cmap=plotting.cm.purple_green, threshold=3)
Add a new overlay on the existing figure
Example: Visualizing a probabilistic atlas: the default mode in the MSDL atlas

plot_markers

display.add_markers(coords, marker_color="y", marker_size=100)
Add seed based MNI coordinates as spheres on top of statistical image or EPI image. This is useful for seed based regions specific interpretation of brain images.
Example: Producing single subject maps of seed-to-voxel correlation

plot_scalebar

display.annotate(scalebar=True)
Adds annotations such as a scale bar, or the cross of the cut coordinates
Example: More plotting tools from nilearn

7.5. Displaying or saving to an image file

To display the figure when running a script, you need to call nilearn.plotting.show (this is just an alias to matplotlib.pyplot.show):

from nilearn import plotting
plotting.show()

The simplest way to output an image file from the plotting functions is to specify the output_file argument:

from nilearn import plotting
plotting.plot_stat_map(img, output_file='pretty_brain.png')

In this case, the display is closed automatically and the plotting function returns None.


The display object returned by the plotting function has a savefig method that can be used to save the plot to an image file:

from nilearn import plotting
display = plotting.plot_stat_map(img)
display.savefig('pretty_brain.png')
# Remember to close the display
display.close()

7.6. Surface plotting

Plotting functions required to plot surface data or statistical maps on a brain surface.

Added in version 0.3.

plot_surf_roi

plot_surf_roi
Plotting surface atlases on a brain surface
Example: Loading and plotting of a cortical surface atlas

plot_surf_stat_map

plot_surf_stat_map
Plotting statistical maps onto a brain surface
Example: Seed-based connectivity on the surface

7.7. Interactive plots

Nilearn also has functions for making interactive plots that can be seen in a web browser. There are two kinds of plots for which an interactive mode is available:

Added in version 0.5: Interactive plotting is new in nilearn 0.5

Added in version 0.9.0: Nilearn offers the possibility to select different plotting engines (either matplotlib or plotly) for most surface plotting functions.

7.7.1. 3D Plots of statistical maps or atlases on the cortical surface

For 3D surface plots of statistical maps or surface atlases, you have different options depending on what you want to do and the packages you have installed.

  • If you have plotly installed: Since Nilearn 0.9.0, it is possible to use plotly as the plotting engine in most surface plotting functions of Nilearn like plot_surf, plot_surf_stat_map, or plot_surf_roi. By default these functions use matplotlib as the plotting engine which results in non-interactive plots. By setting engine='plotly', an interactive version of these plots will be automatically opened in your web browser. In addition, if you have kaleido installed, you can save the plots automatically to png files. Finally, the plotly figure returned can be further customized in order to make more elaborate plots.

  • If you don’t have plotly:

    • If you don’t need to save the plots to png and want to interactively visualize the surface, you can rely on view functions:

      • view_img_on_surf which takes as input a volume statistical map and projects it on the cortical surface before making a 3D interactive plot.

      • view_surf which takes as input a surface map and a cortical mesh and produces a 3D interactive plot on the cortical surface.

    • The last option is to rely on surface plotting functions (plot_surf, plot_surf_stat_map, plot_surf_roi…) with the default matplotlib plotting engine. In this case, you will be able to save to png but you will lose the interactivity of the plots.

7.7.1.1. view_img_on_surf: Surface plot using a 3D statistical map

You can use view_img_on_surf to display a 3D statistical map projected on the cortical surface:

from nilearn import plotting, datasets
img = datasets.fetch_localizer_button_task()['tmap']
view = plotting.view_img_on_surf(img, threshold='90%', surf_mesh='fsaverage')

If you are running a notebook, displaying view will embed an interactive plot (this is the case for all interactive plots produced by nilearn’s “view” functions):

../_images/plotly_surface_plot_notebook_screenshot.png

If you are not using a notebook, you can open the plot in a browser like this:

view.open_in_browser()

This will open this 3D plot in your web browser:

../_images/plotly_surface_plot.png

Or you can save it to an html file:

view.save_as_html("surface_plot.html")

7.7.1.2. view_surf: Surface plot using a surface map and a cortical mesh

You can use view_surf to display a 3D surface statistical map over a cortical mesh:

from nilearn import plotting, datasets
destrieux = datasets.fetch_atlas_surf_destrieux()
fsaverage = datasets.fetch_surf_fsaverage()
view = plotting.view_surf(fsaverage['infl_left'], destrieux['map_left'],
                          cmap='gist_ncar', symmetric_cmap=False)
view.open_in_browser()
../_images/plotly_surface_atlas_plot.png

7.7.1.3. plot_surf_stat_map: Surface plot using a surface map and a cortical mesh

If you have plotly installed, you can also use plot_surf_stat_map with the engine parameter set to “plotly” to display a statistical map over a cortical mesh:

from nilearn import plotting, datasets, surface
fsaverage = datasets.fetch_surf_fsaverage()
motor_images = datasets.fetch_neurovault_motor_task()
mesh = surface.load_surf_mesh(fsaverage.pial_right)
map = surface.vol_to_surf(motor_images.images[0], mesh)
fig = plotting.plot_surf_stat_map(mesh, map, hemi='right',
                                  view='lateral', colorbar=True,
                                  threshold=1.2,
                                  bg_map=fsaverage.sulc_right,
                                  engine='plotly')
fig.show()
../_images/plotly_plot_surf_stat_map.png

7.7.2. 3D Plots of connectomes

For 3D plots of a connectome, use view_connectome. To see only markers, use view_markers.

view_connectome: 3D plot of a connectome:

view = plotting.view_connectome(correlation_matrix, coords, edge_threshold='90%')
view.open_in_browser()
../_images/plotly_connectome_plot.png

7.7.3. 3D Plots of markers

view_markers: showing markers (e.g. seed locations) in 3D:

from nilearn import plotting
dmn_coords = [(0, -52, 18), (-46, -68, 32), (46, -68, 32), (1, 50, -5)]
view = plotting.view_markers(dmn_coords, ['red', 'cyan', 'magenta', 'orange'],
                             marker_size=10)
view.open_in_browser()
../_images/plotly_markers_plot.png

7.7.4. Interactive visualization of statistical map slices

view_img: open stat map in a Brainsprite viewer (https://github.com/simexp/brainsprite.js):

from nilearn import plotting, datasets
img = datasets.fetch_localizer_button_task()['tmap']
html_view = plotting.view_img(img, threshold=2, vmax=4,
                              cut_coords=[-42, -16, 52],
                              title="Motor contrast")

in a Jupyter notebook, if html_view is not requested, the viewer will be inserted in the notebook:

../_images/view_img_screenshot_notebook.png

Or you can open a viewer in your web browser if you are not in a notebook:

html_view.open_in_browser()

Finally, you can also save the viewer as a stand-alone html file:

html_view.save_as_html('viewer.html')