A comparative analysis of image segmentation techniques toward automatic risk prediction of solitary pulmonary nodules

Abstract

Lung cancer is considered as a leading cause of death throughout the globe. Manual interpretation of cancer detection is time consuming and thus increases the death rate. With the help of improvement in medical imaging technology, a computer-aided diagnostics system could be an aid to combat this disease. Automatic segmentation of a region of interest is one of the most challenging problem in medical image analysis. An inaccurate segmentation of solitary pulmonary nodule may lead to an erroneous prediction of the disease. In this paper, we perform a comparative study among the available segmentation techniques, which can automatically segment the solitary pulmonary nodules from high-resolution computed tomography (CT) images and then we propose a computerized lung nodule risk prediction model based on the best segmentation technique.