. Author manuscript; available in PMC: 2005 Nov 14.

Published in final edited form as: IEEE Trans Med Imaging. 2004 Jul;23(7):903–921. doi: 10.1109/TMI.2004.828354

TABLE I.

Comparison of STAPLE and Voting Rule Estimates of Rater Performance. the STAPLE Estimate Correctly Identifies the Expert as Having Superior Performance to That of the Three Medical Students

	Students
Segmentations	1	2	3	expert
Estimate of rater performance from a voting rule.
$\hat{p}$	0.88660	0.91366	0.86856	0.79768
$\hat{q}$	0.99940	0.99922	0.99946	0.99972
STAPLE estimate of rater performance.
$\hat{p}$	0.88095	0.91667	0.88690	1.0
$\hat{q}$	0.99890	0.99872	0.99904	1.0
Exact rater performance from conesnsus segmentation.
p	0.88095	0.91667	0.88690	1.0
q	0.99890	0.99872	0.99904	1.0