Application of neural network based hybrid system for lung nodule detection

Abstract

A hybrid lung nodule detection (HLND) system based on artificial neural network architectures is developed for improving diagnostic accuracy and speed for lung cancerous pulmonary radiology. The configuration of the HLND system includes the following processing phases: data acquisition and pre-processing, in order to reduce and to enhance the figure-background contrast; quick selection of nodule suspects based upon the most prominent feature of nodules, the disc shape; and complete feature space determination and neural classification of nodules. Nodule suspects are captured and stored in 32*32 images after first two processing phases. Eight categories including true nodule, rib-crossing, rib-vessel crossing, end vessel, vessel cluster, bone, rib edge, and vessel are identified for further neural analysis and classification. Extraction of shape features is performed through the edge enhancement self-organized Kohenen feature map, histogram equalization, and evaluation of marginal distribution curves. A supervised back-propagation-trained neural network is developed for recognition of the derived feature curve, a normalized marginal distibution curve.<>