Automated delineation of pulmonary artery cross-sections in CT scans

The purpose of this project was to develop a software tool for studying the dynamic changes of the cross-sections of the pulmonary arteries from computer-tomography (CT) scans. The software for analyzing CT images in the DICOM format was developed using Cocoa, the Apple's object-oriented C++ application programming interface (API) for the Mac OS X operating system, specifically using Mac OS X Lion. The software was developed as a plugin for OsiriX, which is an open-source, widely used DICOM viewer. The main algorithm for detecting the artery border first establishes a seed point within the cross-sectional area of the pulmonary artery, based off of an initial user click. Then the artery border is detected in an automated way by launching 24 radial lines from this seed point. In a similar fashion the border detection continues onto the subsequent frames.