Try to annotate the lesion yourself first: trace a single closed loop, then press Calculate to reveal how two experts did.
All three raters
Majority vote
Mask average
01
STAPLE consensus
01
STAPLE per-rater estimates
Sensitivity is how much of the estimated truth a rater captures; specificity is how well they avoid over-drawing. Trace a careless or over-inclusive contour and watch your row drop.