Automated system-based classification of lung cancer using machine learning

摘要

Lung malignant growth is the well-known reason for death identified due to cancer worldwide. Therefore, to help the radiologist to detect it correctly, automated computer techniques come up with several machine learning classification. For such an automated technique, machine learning algorithms have been applied for the classification of CT scan lung images. This includes two proposed novel features Gabor energy, Gabor entropy, and five grey level co-occurrence matrix (GLCM). These new features are distinct and help in boosting the performance of the classifier to achieve higher accuracy. The proposed method has been simulated on 450 CT scan lung images acquired from the publicly available Lung Image Database Consortium (LIDC) and Image Database Resource Initiative (IDRI) dataset. As a result, the accuracy of 100%, 99%, 83%, and 92% have been achieved from support vector machine (SVM), neural networks (NN), Naive Bayes (NB), and perceptron, respectively.