A 3-D multicriterion automatic segmentation algorithm is developed to impro
ve accuracy of delineation of pulmonary nodules on helical computed tomogra
phy (CT) images by removing their adjacent structures. The algorithm applie
s multiple gray-value thresholds to a nodule region of interest (ROI). At e
ach threshold level, the nodule candidate in the ROI is automatically detec
ted by labeling 3-D connected components, followed by a 3-D morphologic ope
ning operation. Once the nodule candidate is found, its two specific parame
ters, gradient strength of the nodule surface and a 3-D shape compactness f
actor, can be computed. The optimal threshold can be determined by analyzin
g these parameters. Our experiments with in vivo nodules demonstrate the fe
asibility of employing this algorithm to improve the accuracy of nodule del
ineation, especially for small nodules less than 1 cm in diameter. This dis
closes the potential of the algorithm for accurate characterizations of nod
ules (e.g., volume, change in volume over time) at an early stage, which ca
n help to provide valuable guidance for further clinical management. (C) 19
99 Society of Photo-Optical Instrumentation Engineers.