cle.imshow doesnt work properly?

[turkalpmd]

I'm also trying multiplying and diverting with 255 , that not worked.

What is the problem I cant understood.

[haesleinhuepf]

Hi @turkalpmd ,

can you share the code that you used?

Thanks!

Best, Robert

[turkalpmd]

It is fastest response, Thank you.

I am using Colab, I'm studying for some segmentation code . I can sincerely say that you have done very well in segmentation. Thanks for sharing.

Here is the code;

from skimage.io import imread, imshow
import matplotlib.pyplot as plt
import numpy as np

path = "BMP4blastocystC3-cropped_resampled_8bit.tif"
input_image = imread(path)

import pyclesperanto_prototype as cle
# select a specific OpenCL / GPU device and see which one was chosen
cle.select_device('Tesla P100-PCIE-16GB')

# grey value dilation, a.k.a. a maximum-filter
thicker_membranes = cle.maximum_sphere(input_image/255, radius_x=10, radius_y=10)

binary = thicker_membranes < threshold

labels = cle.voronoi_labeling(binary)

fig, axs = plt.subplots(1, 3, figsize=(15, 15))
cle.imshow(thicker_membranes, plot=axs[0])
cle.imshow(binary, plot=axs[1])
cle.imshow(labels, plot=axs[2], labels=True)

[haesleinhuepf]

So the code you pasted doesn't run, because the variable threshold is not defined. Assuming it's 500 produces exactly the output you showed above.

I can make your script work by changing the thresholding step from

binary = thicker_membranes < threshold

to

threshold = 0.5
binary = thicker_membranes > threshold

But I'm not sure if that's the result you're attempting to achieve:

I'd recommend building the workflow step-by-step and looking carefully at intermediate results. E.g. start like this in a jupyter/colab cell:

from skimage.io import imread, imshow
import matplotlib.pyplot as plt
import numpy as np

path = "BMP4blastocystC3-cropped_resampled_8bit.tif"
input_image = imread(path)

import pyclesperanto_prototype as cle
# select a specific OpenCL / GPU device and see which one was chosen
cle.select_device('Tesla P100-PCIE-16GB')

# grey value dilation, a.k.a. a maximum-filter
thicker_membranes = cle.maximum_sphere(input_image/255, radius_x=10, radius_y=10)

You can then use cle.imshow(thicker_membranes) to view the result, or just

thicker_membranes

And it will show this output (I think it should show membranes, but I think it doesn't):

Then, go ahead and apply a threshold:

threshold = 0.5
binary = thicker_membranes > threshold
binary

From here, you may want to use connectect-component labeling:

labels = cle.connected_components_labeling_box(binary)
labels

Let me know if it works for you and have a great weekend!

Best, Robert

[turkalpmd]

Yes, It is worked. Thx you for ultra-fast replies.

[turkalpmd]

Code is runnig