Classification of Primary and Secondary Brain Tumor Using Extreme Learning Machine

Abstract

Brain tumor is a growth of the abnormal tumor cells in the human brain. The brain tumor is classified into two types, namely primary tumor, which starts within the brain, and secondary tumor or known as a brain metastasis tumor. The primary tumors consist of various categories; however, the most common and terminal tumor is Glioblastoma. The difference between the metastasis tumor and the end-of-life phase glioblastoma will be complicated to distinguish if the origin of the tumor cannot be identified. The limitations in differentiating Glioblastoma and Metastasis from MRI images led to the clinical dilemma since both tumors require different further treatments. Therefore, we conducted research to assist and simplify the classification process to identify the primary and secondary brain tumors based on MRI digital image processing. In this study, we proposed a methodology such as Extreme Learning Machine for classification and Gray Level Co-Occurrence Matrix for the feature extraction. The total data used in this study was 100 MRI images with a proportion of 70:30 for training and testing data respectively. The system achieved an accuracy rate of 96.67% in classifying primary and secondary brain tumors.