In this paper we present a computationally efficient segmentation algorithm
for breast masses on sonography that is based on maximizing a utility func
tion over partition margins defined through gray-value thresholding of a pr
eprocessed image. The performance of the segmentation algorithm is evaluate
d on a database of 400 cases in two ways. Of the 400 cases, 124 were comple
x cysts, 182 were benign solid lesions, and 94 were malignant lesions. In t
he first evaluation, the computer-delineated margins were compared to manua
lly delineated margins. At an overlap threshold of 0.40, the segmentation a
lgorithm correctly delineated 94% of the lesions. In the second evaluation,
the performance of our computer-aided diagnosis method on the computer-del
ineated margins was compared to the performance of our method on the manual
ly delineated margins. Round robin evaluation yielded A, values of 0.90 and
0.87 on the manually delineated margins and the computer-delineated margin
s, respectively, in the task of distinguishing between malignant and nonmal
ignant lesions. (C) 2001 American Association of Physicists in Medicine.