A SHIFT-INVARIANT NEURAL-NETWORK FOR THE LUNG FIELD SEGMENTATION IN CHEST RADIOGRAPHY

We have developed a computerized method using a neural network for the segmentation of lung fields in chest radiography. The lung is the pri mary region of interest in routine chest radiography diagnosis. Since computer is expected to perform disease pattern search automatically, it is important to design appropriate algorithms to delineate the regi on of interest. A reliable segmentation method is essential to facilit ate subsequent searches for image patterns associated with lung diseas es. In this study, we employed a shift invariant neural network couple d with error back-propagation training method to extract the lung fiel ds. A set of computer algorithms were also developed for smoothing the initially detected edges of lung fields. Our preliminary results indi cated that 86% of the segmented lung fields globally matched the origi nal chest radiographs. We also found that the method facilitates the d evelopment of computer algorithms in the field of computer-aided diagn osis.