Simple volume slicing

Simple volume slicing

Here we present an example for visualizing slices from 3D images.

from __future__ import division

import os
import nibabel as nib
from dipy.data import fetch_bundles_2_subjects
from dipy.viz import window, actor, ui

Let’s download and load a T1.

fetch_bundles_2_subjects()

fname_t1 = os.path.join(os.path.expanduser('~'), '.dipy',
                        'exp_bundles_and_maps', 'bundles_2_subjects',
                        'subj_1', 't1_warped.nii.gz')


img = nib.load(fname_t1)
data = img.get_data()
affine = img.affine

Create a Renderer object which holds all the actors which we want to visualize.

renderer = window.Renderer()
renderer.background((0.5, 0.5, 0.5))

Render slices from T1 with a specific value range

The T1 has usually a higher range of values than what can be visualized in an image. We can set the range that we would like to see.

mean, std = data[data > 0].mean(), data[data > 0].std()
value_range = (mean - 0.5 * std, mean + 1.5 * std)

The slice function will read data and resample the data using an affine transformation matrix. The default behavior of this function is to show the middle slice of the last dimension of the resampled data.

slice_actor = actor.slicer(data, affine, value_range)

The slice_actor contains an axial slice.

renderer.add(slice_actor)

The same actor can show any different slice from the given data using its display function. However, if we want to show multiple slices we need to copy the actor first.

slice_actor2 = slice_actor.copy()

Now we have a new slice_actor which displays the middle slice of sagittal plane.

slice_actor2.display(slice_actor2.shape[0]//2, None, None)

renderer.add(slice_actor2)

renderer.reset_camera()
renderer.zoom(1.4)

In order to interact with the data you will need to uncomment the line below.

# window.show(renderer, size=(600, 600), reset_camera=False)

Otherwise, you can save a screenshot using the following command.

window.record(renderer, out_path='slices.png', size=(600, 600),
              reset_camera=False)
../../_images/slices.png

Simple slice viewer.

Render slices from FA with your colormap

It is also possible to set the colormap of your preference. Here we are loading an FA image and showing it in a non-standard way using an HSV colormap.

fname_fa = os.path.join(os.path.expanduser('~'), '.dipy',
                        'exp_bundles_and_maps', 'bundles_2_subjects',
                        'subj_1', 'fa_1x1x1.nii.gz')

img = nib.load(fname_fa)
fa = img.get_data()

Notice here how the scale range is (0, 255) and not (0, 1) which is the usual range of FA values.

lut = actor.colormap_lookup_table(scale_range=(0, 255),
                                  hue_range=(0.4, 1.),
                                  saturation_range=(1, 1.),
                                  value_range=(0., 1.))

This is because the lookup table is applied in the slice after interpolating to (0, 255).

fa_actor = actor.slicer(fa, affine, lookup_colormap=lut)

renderer.clear()
renderer.add(fa_actor)

renderer.reset_camera()
renderer.zoom(1.4)

# window.show(renderer, size=(600, 600), reset_camera=False)

window.record(renderer, out_path='slices_lut.png', size=(600, 600),
              reset_camera=False)
../../_images/slices_lut.png

Simple slice viewer with an HSV colormap.

Now we would like to add the ability to click on a voxel and show its value on a panel in the window. The panel is a UI element which requires access to different areas of the visualization pipeline and therefore we don’t recommend using it with window.show. The more appropriate way is to use the ShowManager object, which allows accessing the pipeline in different areas.

show_m = window.ShowManager(renderer, size=(1200, 900))
show_m.initialize()

We’ll start by creating the panel and adding it to the ShowManager

label_position = ui.TextBlock2D(text='Position:')
label_value = ui.TextBlock2D(text='Value:')

result_position = ui.TextBlock2D(text='')
result_value = ui.TextBlock2D(text='')

panel_picking = ui.Panel2D(size=(250, 125),
                           position=(20, 20),
                           color=(0, 0, 0),
                           opacity=0.75,
                           align="left")

panel_picking.add_element(label_position, (0.1, 0.55))
panel_picking.add_element(label_value, (0.1, 0.25))

panel_picking.add_element(result_position, (0.45, 0.55))
panel_picking.add_element(result_value, (0.45, 0.25))

renderer.add(panel_picking)

Add a left-click callback to the slicer. Also disable interpolation so you can see what you are picking.

def left_click_callback(obj, ev):
    """Get the value of the clicked voxel and show it in the panel."""
    event_pos = show_m.iren.GetEventPosition()

    obj.picker.Pick(event_pos[0],
                    event_pos[1],
                    0,
                    renderer)

    i, j, k = obj.picker.GetPointIJK()
    result_position.message = '({}, {}, {})'.format(str(i), str(j), str(k))
    result_value.message = '%.8f' % data[i, j, k]


fa_actor.SetInterpolate(False)
fa_actor.AddObserver('LeftButtonPressEvent', left_click_callback, 1.0)

# show_m.start()

Create a mosaic

By using the copy and display method of the slice_actor becomes easy and efficient to create a mosaic of all the slices.

So, let’s clear the renderer and change the projection from perspective to parallel. We’ll also need a new show manager and an associated callback.

renderer.clear()
renderer.projection('parallel')

result_position.message = ''
result_value.message = ''

show_m_mosaic = window.ShowManager(renderer, size=(1200, 900))
show_m_mosaic.initialize()


def left_click_callback_mosaic(obj, ev):
    """Get the value of the clicked voxel and show it in the panel."""
    event_pos = show_m_mosaic.iren.GetEventPosition()

    obj.picker.Pick(event_pos[0],
                    event_pos[1],
                    0,
                    renderer)

    i, j, k = obj.picker.GetPointIJK()
    result_position.message = '({}, {}, {})'.format(str(i), str(j), str(k))
    result_value.message = '%.8f' % data[i, j, k]

Now we need to create two nested for loops which will set the positions of the grid of the mosaic and add the new actors to the renderer. We are going to use 15 columns and 10 rows but you can adjust those with your datasets.

cnt = 0

X, Y, Z = slice_actor.shape[:3]

rows = 10
cols = 15
border = 10

for j in range(rows):
    for i in range(cols):
        slice_mosaic = slice_actor.copy()
        slice_mosaic.display(None, None, cnt)
        slice_mosaic.SetPosition((X + border) * i,
                                 0.5 * cols * (Y + border) - (Y + border) * j,
                                 0)
        slice_mosaic.SetInterpolate(False)
        slice_mosaic.AddObserver('LeftButtonPressEvent',
                                 left_click_callback_mosaic,
                                 1.0)
        renderer.add(slice_mosaic)
        cnt += 1
        if cnt > Z:
            break
    if cnt > Z:
        break

renderer.reset_camera()
renderer.zoom(1.0)

# show_m_mosaic.ren.add(panel_picking)
# show_m_mosaic.start()

If you uncomment the two lines above, you will be able to move the mosaic up/down and left/right using the middle mouse button drag, zoom in/out using the scroll wheel, and pick voxels with left click.

window.record(renderer, out_path='mosaic.png', size=(900, 600),
              reset_camera=False)
../../_images/mosaic.png

A mosaic of all the slices in the T1 volume.

Example source code

You can download the full source code of this example. This same script is also included in the dipy source distribution under the doc/examples/ directory.