Automatic Detection of Lung Cancer Nodules in Computerized Tomography Images

Abstract

Lung cancer is the prime factor in cancer related deaths due to increasing rate of smoking and air pollution. The survival rate of cancer patients increases to 52% if it is localized, and decreases to 4% if it is metastasized. The existing system use simple thresholding approach and pattern recognition method to segment lung and identify nodules in the lung computerized tomography (CT) images. But quality of this process is affected by the image acquisition protocols, scanner types, and inhomogeneity of intensities in the lung region. In order to overcome the drawbacks of the previous process and for separation of nodules from non-nodules, there is a need for an automatic computer aided detection system. This paper proposes an automatic method for detecting nodules from patient lung CT images. In this automatic approach first step is a lung segmentation to differentiate lung and background and make the detection of nodule simple. Then the lung region is segmented by Contextual Clustering based region growing method. Vector quantization method is used to identify the nodules present in the lungs. False positive reduction is done by using various combination of rule based filtering, support vector machine, k-Nearest Neighbours and Random Forest classifier. The performance of this method is evaluated in terms of accuracy, sensitivity and specificity.