Slicing and cropping#

When working with larger image data, it often makes sense to crop out regions and focus on them for further analysis. For cropping images, we use the same “:”-syntax, we used when indexing in lists and exploring multi-dimensional image data.

import numpy as np
from skimage.io import imread, imshow

We start by loading a 3D image and printing its size.

image = imread("../../data/Haase_MRT_tfl3d1.tif")

image.shape

(192, 256, 256)

Slicing#

For visualizing 3D images using scikit-image’s imshow, we need to select a slice to visualize. For example, a Z-slice:

slice_image = image[100]

imshow(slice_image)

<matplotlib.image.AxesImage at 0x2b54f73d340>
../_images/b9ec53895c9942461dd35e1b88b95a9557441ceb09c074bd865d8bb6ddfdb74b.png

We can also select a plane where all pixels have the same Y-position. We just need to specify, that we would like to keep all pixels in Z using the : syntax.

slice_image = image[:, 100]

imshow(slice_image)

<matplotlib.image.AxesImage at 0x2b54f836af0>
../_images/ad292cbffaab643c3d9273781ac2d7a8cb9aeed56d351d1c201b74c79fcbef31.png

Cropping#

We can also select a sub-stack using indexing in the square brackets.

sub_stack = image[50:150]

sub_stack.shape

(100, 256, 256)

We can also select a sub-region in X. If we want to keep all pixels along Z and Y (the first two dimensions), we just specify : to keep all.

sub_region_x = image[:, :, 100:200]

imshow(sub_region_x[100])

<matplotlib.image.AxesImage at 0x2b54f8ae850>
../_images/38728fe7d44860c0fac1767af8007785db40a1042ac2162ae73b3055cddc2023.png

For selectinng all pixels in one direction above a given value, we just need to specify the start before :.

sub_region_y = image[:, 100:]

imshow(sub_region_y[100])

<matplotlib.image.AxesImage at 0x2b54f90ae20>
../_images/97bd447a765eff73cd22753868fe6c2e3f250a63db6ed468f116ff74bd9615b7.png

Similarly, we can select all pixels up to a given position.

sub_region_x2 = image[:, :, :50]

imshow(sub_region_x2[100])

<matplotlib.image.AxesImage at 0x2b550943d30>
../_images/b1b6f01794a399ac2db4297973bb8e34233be52f6b5d98b70d683d3a5cd5f8df.png

Last but not least, this is how a cropped cube is specified.

cropped_cube = image[80:130, 120:170, :50]

cropped_cube.shape

(50, 50, 50)

imshow(cropped_cube[20])

<matplotlib.image.AxesImage at 0x2b5509a4a00>
../_images/474a83bfd6b87d735ccfa7e5e30c1fdae98c834ef6ced02f2a06bc9f939c8ef0.png

And this is how a maximum-intensity projections of this cropped cube look like.

maximum_intensity_projection_along_z = np.max(cropped_cube, axis=0) 
imshow(maximum_intensity_projection_along_z)

<matplotlib.image.AxesImage at 0x2b550a03f40>
../_images/b60a8d4d3c16d68ffeb90787bbe80d7455732026daee8c574df5a01ff6518f14.png 
maximum_intensity_projection_along_y = np.max(cropped_cube, axis=1) 
imshow(maximum_intensity_projection_along_y)

<matplotlib.image.AxesImage at 0x2b550a5e8e0>
../_images/6a8879bc26fa9b800fb1edf6cccbd641a2638dd0796f9c417c3015f452f385ce.png