Have copy pasted the metrics.py file and commented out everything not necessary to reproduce said issue. Please let me know if I made a mistake somewhere.
But TLDR; is the size of the (isolated) False Positive Punished in this (lesion-wise dice) metric? If not, Why?
Simple Reproducible Example: https://colab.research.google.com/drive/1vhVSYp5PnRE989CXa5zsLnfNZLCTR6p_?usp=sharing
Have copy pasted the metrics.py file and commented out everything not necessary to reproduce said issue. Please let me know if I made a mistake somewhere.
But TLDR; is the size of the (isolated) False Positive Punished in this (lesion-wise dice) metric? If not, Why?