A Hybrid Lung Nodule Detection (HLND) system is developed for improving detection accuracy and speed for lung cancerous pulmonary radiology. The configuration of the HLND system includes (i) pre-processing, in order to enhance the figure-background contrast based upon the energy content of nodule; (ii) quick selection of nodule suspects based upon the most prominent feature of nodules, the disc shape; and (iii) complete feature space determination and neural network classification of nodules, based on the characteristics of edges in nodules. Nodule suspects are captured and stored in 32 x 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. A supervised back propagation artificial neural network architecture is developed for training and recognition of feature curves of nodules. Results show that this detection system is able to identify true nodule at accuracy up to 93% with false detection reduced down to 7%.
