-
-
Notifications
You must be signed in to change notification settings - Fork 53
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
RandomTiler issue with tissue_percent parameter #611
Comments
Hi @Neuro-nerd-scientist thank you for opening this issue! The behavior you show is indeed strange, and it might be related to how uniform the image is in the center. Would you be able to put this image somewhere online so we can have a look and test a few things? |
@Neuro-nerd-scientist after some investigation it seems like it's related to #360, but probably adding a border of 10px is not enough in your case, let me work on a fix :) |
Thank you @alessiamarcolini I'll take a look at the post you mentioned. Please keep me updated, I really appreciate it :) |
hey @Neuro-nerd-scientist can we integrate your image in our testing suite and add specific use cases for that? this will help us ensure we are not introducing regressions in the future. |
The image is from the Allen Institute's Mouse Neuroanatomy Database - it's open access- but this is a sample from one of their brain section images. Is there anything I can do either with regards to the image, or with histolab to account for the behaviour I'm experiencing? |
Hello,
I'm working with histolab's RandomTiler to extract some sample tiles from an image of a larger sample of tissue. I have successfully loaded my image (jpg format) into python using PIL and then initialise the image as a Slide object using Histolab's Slide class
I notice that when I use the default tissue_percent value (80%), all the tiles are aggregated on the top of the sample - this behaviour continues for high tissue_percent values as well (tissue_percent>50). To get random sampling throughout the entirety of my tissue sample, I need to supply very low tissue_percent values (tissue_percent<5%) to get uniform sampling throughout the tissue. I have included three screenshots - showing different tissue_percent values with check_tissue set to True to show the odd tiling patterns this is producing:
Example 1 - tissue_percent set to 80% with check_tissue set to True
Example 2 - tissue_percent set to 20% with check_tissue set to True
This result was unexpected but also anomalous. Because again the tiles are not being extracted throughout the entirety of the tissue ( which I get at extremely low tissue_percent values)
Example 3 - tissue_percent set to 1% with check_tissue set to True
I'm not sure why this is occurring, and why I need such low tissue_percent values to be supplied to the RandomTiler extractor to get tiles sampled throughout my whole image. If anyone has experienced this, or has a solution to this issue, please let me know. Any assistance is greatly appreciated.
Thank you for your time, efforts and consideration.
The text was updated successfully, but these errors were encountered: