Using regionprops in Python

Question

I am trying to analyze greyscale TIFF stacks, in which a given frame will look like this. I filter it (using Gaussian blur), and then binarize it (using Otsu's method for threshold).

MATLAB code, which works great:

image_conncomp = bwconncomp(image_binary); # entire stack is held in image_binary

for i=1:image_conncomp.NumObjects
    object_size = length(image_conncomp.PixelIdxList{i});
end

Each white spot in the example image is picked up, and its volume (in pixels) is pretty accurately given by object_size.

Python code:

from skimage import measure

labels = measure.label(image_binary, background=1) # same image_binary as above
propsa = measure.regionprops(labels)

for label in propsa:
    object_size = len(label.coords)

The Python code seems to work decently... except that most detected objects will have object_size of 1 - 200, and then a couple will have a size of several thousand pixels.

What are these functions doing differently? I would be happy to try another approach in Python to get calculate object sizes, but I struggled to find another one. It'd be great to have a Python version of this code, if I could find a good substitute for Matlab's bwconncomp function.

I honestly don't understand the output in its raw form. It's supposed to label the objects that it recognizes, but not sure exactly how it does it — oofin, Jul 09 '18 at 16:41
Why did you set background=1? Surely that is not the default in Matlab? Also, the RegionProp objects have some nice properties, including area which is what you want: `for prop in propsa: object_size = prop.area`. Anyway, I suggest using skimage.color.label2rgb to look at the label image and make sure that that step is working. The rest looks good (assuming the binaries really are identical, and that the background is set to 0, I think) — Juan, Jul 10 '18 at 04:27

score 4 · Answer 1 · answered Jul 21 '18 at 11:40

4

Something like this?

from skimage.io import imread, imshow
from skimage.filters import gaussian, threshold_otsu
from skimage import measure
import matplotlib.pyplot as plt

original = imread('https://i.stack.imgur.com/nkQpj.png')
blurred = gaussian(original, sigma=.8)
binary = blurred > threshold_otsu(blurred)
labels = measure.label(binary)

plots = {'Original': original, 'Blurred': blurred, 
         'Binary': binary, 'Labels': labels}
fig, ax = plt.subplots(1, len(plots))
for n, (title, img) in enumerate(plots.items()):
    cmap = plt.cm.gnuplot if n == len(plots) - 1 else plt.cm.gray
    ax[n].imshow(img, cmap=cmap)
    ax[n].axis('off')
    ax[n].set_title(title)
plt.show(fig)

props = measure.regionprops(labels)
for prop in props:
    print('Label: {} >> Object size: {}'.format(prop.label, prop.area))

Output:

Label: 1 >> Object size: 37
Label: 2 >> Object size: 66
Label: 3 >> Object size: 1

answered Jul 21 '18 at 11:40

Tonechas

11,520
14
37
68

1

This works great for measuring area. However, I'm looking for volume. Generally, a stack will have 34 frames similar to the sample image above. Objects will appear in more than one frame, but not all - so I have to look at the entire stack all at the same time to actually get volume. To measure volume, I just want to count the # of pixels found in the object – oofin Jul 21 '18 at 17:25
In that case you need to loop over the stack and compute the volume as the sum of the areas of the regions which correspond to the same object. – Tonechas Jul 21 '18 at 17:47
but how do i know that they are the same object? the only method I can think of is to compute the distance between centroids, but that sounds very time expensive – oofin Jul 21 '18 at 18:10
I added the stack (converted to png, Chrome can't render .tiff's) to the bottom of original question – oofin Jul 21 '18 at 20:06
I am not able to download the stack from the url you provided. Could you post a Dropbox/Google drive link? – Tonechas Jul 21 '18 at 21:51
I missed this discussion. @oofin, the property "area" actually returns a volume if your image is 3D – Juan Feb 02 '19 at 04:53

Sandipan Dey · Answer 2 · 2019-11-05T03:51:19.873

We could do the same by first applying scipy.ndimage's morphological closing on the thresholded binary image followed by the label() function to merge the connected regions in the binary image, as shown below (size of the regions are a bit different though and will depend upon the size of the morphological kernel):

from scipy.ndimage import label
from scipy.ndimage.morphology import binary_closing
from skimage.filters import threshold_otsu
import matplotlib.pylab as plt

original = plt.imread('https://i.stack.imgur.com/nkQpj.png')
thres = threshold_otsu(original)
binary = original > thres
binary_closed = binary_closing(binary, structure=np.ones((2,2)))
labeled_image, num_features = label(binary_closed)
feature_areas = np.bincount(labeled_image.ravel())[1:]   
print(feature_areas) # if we use 3x3 SE we shall get two regions with areas 24, 46
# [24 42  1]

plt.figure(figsize=(8,7))
plt.gray()
plt.subplot(221), plt.imshow(original), plt.axis('off'), plt.title('original')
plt.subplot(222), plt.imshow(binary), plt.axis('off'), plt.title('binray')
plt.subplot(223), plt.imshow(binary_closed), plt.axis('off'), plt.title('binray closed')
plt.subplot(224), plt.imshow(labeled_image, cmap='inferno'), plt.axis('off'), plt.title('labelled')
plt.show()

to obtain the following output:

Using regionprops in Python

2 Answers2

Linked

Related