squidpy
squidpy copied to clipboard
scverse
H&E Image Features
Hi, I am currently learning Squidpy calculate image features. I am wondering what is the interpretation of the pattern? I could not figure out why there are circles of low-mean image features(how to interpret the location of the circles). Also, channel 0,1 and 2 do not seem be too different on the tissue sample I am working on.
For this, I used the tutorial on this site https://squidpy.readthedocs.io/en/stable/auto_examples/image/compute_features.html#sphx-glr-auto-examples-image-compute-features-py
I tried custom_features as well, but did not get any difference.
I would be grateful if you can provide some insights of the results I am getting!
hi @Charlottehero sorry completely missed this issue. looks interesting, can you share the actual line of code you used? it maybe computing features on different layers?
Thank you for the reply!
I first loaded in the tif image file our team has using image container function,
img = sq.im.ImageContainer("path/to/image.tif", scale = scale)
and I used the default image feature calculation function in the tutorial.
feature_name = "features_summary" sq.im.calculate_image_features( adata, img, features="summary", key_added=feature_name, n_jobs=4, scale=scale, )
Btw, we are also not sure why the intensity value range is very narrow across the whole tissue. Is there anything wrong with our code?
apologies for late reply, I must say it's difficult to debug with this info. Are you sure that image features calculcation is computed on the same (or a scaled version) of the tissue image that you use for plotting?
Re intesnity ranges, it looks like the iamge is in 8but instead of 16bit so it's not necessarily low (thje max is 255)
@DaisyCuttie I'll close this issue due to inactivity, let me know if the problem persist and feel free to reopen the issue!