3D and 4D niimgs: handling and visualizing

Here we discover how to work with 3D and 4D niimgs.

Downloading tutorial datasets from Internet

Nilearn comes with functions that download public data from Internet

Let’s first check where the data is downloaded on our disk:

from nilearn import datasets

print(f"Datasets are stored in: {datasets.get_data_dirs()!r}")

Datasets are stored in: ['/home/runner/work/nilearn/nilearn/nilearn_data', '/home/runner/nilearn_data']

Let’s now retrieve a motor contrast from a Neurovault repository

motor_images = datasets.fetch_neurovault_motor_task()
motor_images.images

['/home/runner/work/nilearn/nilearn/nilearn_data/neurovault/collection_658/image_10426.nii.gz']

motor_images is a list of filenames. We need to take the first one

tmap_filename = motor_images.images[0]

Visualizing a 3D file

The file contains a 3D volume, we can easily visualize it as a statistical map:

from nilearn import plotting

plotting.plot_stat_map(tmap_filename)

<nilearn.plotting.displays._slicers.OrthoSlicer object at 0x7f853950ef30>

Visualizing works better with a threshold

plotting.plot_stat_map(tmap_filename, threshold=3)

<nilearn.plotting.displays._slicers.OrthoSlicer object at 0x7f852d1c14f0>

Visualizing one volume in a 4D file

We can download resting-state networks from the Smith 2009 study on correspondence between rest and task

rsn = datasets.fetch_atlas_smith_2009(resting=True, dimension=10)["maps"]
rsn

Downloading data from https://www.fmrib.ox.ac.uk/datasets/brainmap+rsns/PNAS_Smith09_rsn10.nii.gz ...

Downloaded 3039232 of 7565016 bytes (40.2%,    1.5s remaining) ...done. (3 seconds, 0 min)

'/home/runner/work/nilearn/nilearn/nilearn_data/smith_2009/PNAS_Smith09_rsn10.nii.gz'

It is a 4D nifti file. We load it into the memory to print its shape.

from nilearn import image

print(image.load_img(rsn).shape)

(91, 109, 91, 10)

We can retrieve the first volume (note that Python indexing starts at 0):

first_rsn = image.index_img(rsn, 0)
print(first_rsn.shape)

(91, 109, 91)

first_rsn is a 3D image.

We can then plot it

plotting.plot_stat_map(first_rsn)

<nilearn.plotting.displays._slicers.OrthoSlicer object at 0x7f85395a1eb0>

Looping on all volumes in a 4D file

If we want to plot all the volumes in this 4D file, we can use iter_img to loop on them.

Then we give a few arguments to plot_stat_map in order to have a more compact display.

for img in image.iter_img(rsn):
    # img is now an in-memory 3D img
    plotting.plot_stat_map(
        img, threshold=3, display_mode="z", cut_coords=1, colorbar=False
    )

• 
• 
• 
• 
• 
• 
• 
• 
• 
• 

Looping through selected volumes in a 4D file

If we want to plot selected volumes in this 4D file, we can use index_img with the slice constructor to select the desired volumes.

Afterwards, we’ll use iter_img to loop through them following the same formula as before.

selected_volumes = image.index_img(rsn, slice(3, 5))

If you’re new to Python, one thing to note is that the slice constructor uses 0-based indexing. You can confirm this by matching these slices to the previous plot above.

for img in image.iter_img(selected_volumes):
    plotting.plot_stat_map(img)

• 
• 

plotting.show is useful to force the display of figures when running outside IPython

plotting.show()

---

To recap, neuroimaging images (niimgs as we call them) come in different flavors:

• 3D images, containing only one brain volume

• 4D images, containing multiple brain volumes.

More details about the input formats in nilearn for 3D and 4D images is given in the documentation section: Inputing data: file names or image objects.

Functions accept either 3D or 4D images, and we need to use on the one hand nilearn.image.index_img or nilearn.image.iter_img to break down 4D images into 3D images, and on the other hand nilearn.image.concat_imgs to group a list of 3D images into a 4D image.

Estimated memory usage: 215 MB