Diagnosis of lung nodule using reinforcement learning and geometric measures

Abstract

This paper uses a set of 3D geometric measures with the purpose of characterizing lung nodules as malignant or benign. Based on a sample of 36 nodules, 29 benign and 7 malignant, these measures are analyzed with a technique for classification and analysis called reforcement learning. We have concluded that this techiniqne allows good discrimination from benign to malignant nodules. © Springer-Verlag Berlin Heidelberg 2005.