Application of Artificial Neural Networks to Chest Image Classification

In this paper the use of a neural network (NN) as a tool for automatic diagnosis of chest pathologies is presented. Images fed to the network are first preprocessed in order to automatically isolate the lung cavities and partially remove the influence of various displacements. The results obtained so far for a group of patients with several pathologies show the feasibility of an NN-based system as an aid for diagnosis in a Radiology Service or, with an exhaustive training over a wider number of pathologies, as an effective automatic diagnosis system.