You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hello, your work is quite outstanding, and I have gained a lot of help from it.
However, I still have some questions about the image preprocessing operations before obtaining the patch-level features that I would like to consult with you.
When preprocessing the input image, after using the OTSU algorithm distinguish between the foreground and background.
And then, when dividing the image into tiles of size 256*256, is it only the tiles that are entirely composed of foreground pixels that are selected?
If one tile contains any background pixel, would it be discarded?
The text was updated successfully, but these errors were encountered:
Yes, it is essential to discard background and non-informative tiles, at least for the sake of computational efficiency. Typically, a simple color filter is sufficient for this purpose.
Nowadays, some handy off-the-shelf tools are available. Personally, I switch between tools like histolab and HistomicsTK.
Hello, your work is quite outstanding, and I have gained a lot of help from it.
However, I still have some questions about the image preprocessing operations before obtaining the patch-level features that I would like to consult with you.
When preprocessing the input image, after using the OTSU algorithm distinguish between the foreground and background.
And then, when dividing the image into tiles of size 256*256, is it only the tiles that are entirely composed of foreground pixels that are selected?
If one tile contains any background pixel, would it be discarded?
The text was updated successfully, but these errors were encountered: