Automatic Lung nodule segmentation and classification in CT images based on SVM

Early detection of lung cancer is of vital importance to successful treatment where Computed Tomography (CT) screening are considered one of the best methods for detection the early signs of lung cancer. Standard Computer Aided Diagnosis (CAD) systems for Lung cancer detection should employ four steps: preprocessing, lungs parenchyma segmentation, nodule detection and reduction of False Positives (FP). In the proposed approach during the preprocessing step, several masks are calculated using thresholding technique and morphological operations, eliminating this way, background and surrounding tissue. Following, suspicious Regions of Interest (ROI) are calculated using a priori information and Hounsfield Units (HU). During feature extraction, numerous features are calculated in order to restrict the suspicious zones. Finally, Support Vector Machine (SVM) algorithm is employed in classification stage.