Classification of lung nodules with feature extraction using CT scan images

Abstract

OBJECTIVE: The main aim is to differentiate the various types of lung nodules using the SVM classifier. By identifying the lung nodules, the cause of lung cancer can be avoided. METHODOLOGY: The major contributions in this system are (i) Patch based division, to partition the original images (ii) Feature extraction stage, to extract feature information (iii) Classification stage, to classify the four types of lung nodules with the help of SVM classifier with pLSA. FINDINGS: This system has an improvement with the Local Tetra Pattern (LTrP) to provide more feature information. This pattern extracts feature information from more than two direction to give accurate results. IMPROVEMENT: This system can be improved with different classifier to achieve accurate classification.