Performance of GLCM Algorithm for Extracting Features to Differentiate Normal and Abnormal Brain Images

Abstract

Brain cancer is a malignant brain tumor that can spread quickly to other parts of the brain and spine. However, not all tumors are malignant and can be treated before they become malignant. The purpose of this study is to discover brain abnormalities based on CT scan images by using T-test algorithm. Thus, it can be one of solution for early detection of brain abnormalities in order to treat it before it becomes a malignant tumor or cancer. As dataset, this research using 40 images consisting of 20 normal brain images and 20 abnormal brain images. There are two algorithms which are used in this research i.e. Gray level co-occurrences matrix (GLCM) for feature extraction and T-Test for brain image classification. Prior to feature extraction, brain image is converted to Graycomatrix in order to adjust the brightness of the image. The final step is image classification by using the T-test algorithm. From 40 test results which are used in this study, GLCM method can extract 8 features that can significantly distinguish the image of normal brain and abnormal brain. For the T-test algorithm, it is found that each feature has a P-value <0.05 which means that extracted features can be used for the further classification process of brain image abnormality. Thus, it can be inferred that this research framework which is employed the GLCM and T-test algorithm can be used to assist the process of early diagnosis of brain cancer.