Athena Hosseinpour - Department of Computer, Islamic Azad University, Guilan, Iran
Farshid Mehrdoust - Department of Computer, Islamic Azad University, Guilan, Iran
Mohsen salehi - Department of Computer, Imam Reza University, Mashhad, Iran

Computer-aided detection (CAD) of pulmonary nodules is critical to assisting radiologists in early identification of lung cancer from computed tomography (CT) scans. This paper describes a Computer-Aided Diagnosis (CAD) system for automatic pulmonary nodules detection on serial CT scans based on Support vector machine classification (SVM) scheme which the size of the nodules are greater than 8mm. Compared with the simple thresholding approach, the SVM yields a more accurate segmentation of the lungs from the chest volume. to identifying initial nodule candidates within the lungs, the proposed system proves to be effective for initial nodule candidates detection and segmentation, as well as existing approaches. False-positive is reduced by rule-based filtering operations in combination with a feature-based support SVM classifier. The proposed system was validated on 47 patient cases from the publically available on-line LIDC (Lung Image Database consortium) database. Experimental results show that our system obtained an overall sensitivity of 89.9 % at a specificity of 4.0 FPs/scan. With respect to comparable previous system, the proposed system shows outperformance and demonstrates its potential for best detection of pulmonary nodules via CT imaging

Computer-aided detection, CT imaging, false-positive rate, Solitary Pulmonary Nodules, support vector machine