An Efficient Classification Method Using GLCM and Decision Tree Classifier

Abstract

The treatment of lung cancer and related diseases is a major concern of medical science. Death due to lung cancer is seen as a leading cause and therefore early detection and treatment becomes a critical necessity. A commonly used technique of imaging to diagnose lung cancer is computed tomography (CT). In disease diagnosis, envisaging techniques play an important role. These methods help detect abnormal tissue-tumor patterns in targeted cancer cells. According to the World Health Organization, among all cancers, lung cancer accounts for around 14 percent. Paper contributes the different five stages as a proposed methodology. Firstly, database like normal lung and disease lung cancers images are collected. Median filter is used for pre-processing. Edges are preserves correctly by the median filter, so it is preferring, i.e., preservation of sharp features. In the third stage, segmentation on the target image is performed to ascertain and isolate the desired cancerous entity from the background. In stage four, features such as area, contraction, energy, entropy and homogeneity are mined using Gray Level Co-occurrence Matrix (GLCM). High demarcation precision and lower processing speed can be achieved by using GLCM. The fifth stage, these mined features are assigned to 5 different classifiers to classify lung cancer from normal lung. A decision tree (DT) classifier achieved highest accuracy for detection and lung cancer classification.