According to this article:

Dice coefficient=

The 2-class DSC variant for class c is expressed in Equation 1, where gic ∈ {0, 1} and pic ∈ [0, 1] represent the ground truth label and the predicted label, respectively. The total number of pixels in an image is denoted by N. The epsilon provides numerical stability to prevent division by zero.

enter image description here

And the author said:

With highly imbalanced data and small ROIs such as skin lesions, FN detections need to be weighted higher than FPs to improve recall rate.

I can't understand why does dice score weigh FP and FN. Could you help me to understand this?

  • $\begingroup$ is it possible it has to do with label indexes, ie negative has zero as ground truth, while positive has one as ground truth?? $\endgroup$ – Nikos M. Jul 26 '20 at 13:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.