SYSTEMS FOR COMPUTER AIDED LUNG NODULE DETECTION IN CHEST TOMOSYNTHESIS IMAGING

摘要

A procedure for image segmentation of a lung in tomosynthesis images includes determining a focal plane image of a lung from among a plurality of tomosynthesis images, determining boundaries of the lung in the focal plane image based on a sequence of best- path algorithms cascaded together, assembling the tomosynthesis images to obtain a 3D image of the lung, determining a boundary of a rib in the 3D image of the lung, and segmenting the lung based on the boundaries of the lung and the boundary of the rib. A procedure for detecting nodules in tomosynthesis images includes generating a blurred nodule template, generating a blurred vessel template and a blurred rib template, determining, based on the blurred nodule template, a nodule candidate in 3D image of a lung, and determining, based on the blurred vessel template and a blurred rib template, that the nodule candidate is a nodule.