Enhanced detecting system for computer-aided diagnosis of CT lung cancer

Abstract

According to The American Cancer Society, In the US. they estimated that there will be 160,390 deaths from lung cancer and that 213,380 new cases will be diagnosed in 2007. The accurate diagnosis of the lung nodule in case of determine either malignant or benign requires a lot of resources to read the ever increasing volume of detecting nodules on high resolution computed tomography (HRCT). So, the motivation for developing and enhancing the performance and accuracy of computer-aided-diagnosis systems (CADx) for HRCT has been the focus of many research groups to alleviate this burden. In this work, we report the results of a new CADx system that provides enhanced detecting performance. We utilize an optimized set of texture features using feature selection and apply the new system using Datasets provided by National Cancer Institute (NCI)’s and The Cancer Imaging Archive (TCIA). The results compared well to previous work with classification accuracy performance of 89.4% and 86% sensitivity and 93% specificity. The implementation details and analysis of the proposed system are described in this paper. The results of the new system will contribute a significant impact on the accuracy of such systems and hence the enhancement of detecting process outcome.