Brain Tumor Classification based on Feature Extraction and Hybridize Regularized Extreme Learning Machine and Sine Cosine Algorithm (SCA-RELM)

Rohit Agarwal and Diwakar Bhardwaj

Knowledge and experience of physician defines the process of cancer treatment. Due to this, physicians and radiologists are assisted by a tumor detection technique which is automatic for detecting brain tumors. In this research, accurate classification of brain tumor is performed by developing a hybrid feature extraction method which includes regularized extreme learning machine techniques. Initially, a Z-score normalization approach is used as a pre-processing process to normalize the brain tumour dataset. After that, approach starts by extracting features from brain images using the hybrid feature extraction method; then, computing the covariance matrix of these features to project them into a new significant set of features using principle component analysis (PCA). Regularized extreme learning machine and sine cosine algorithm (SCA-RELM) are hybridized for classification. This method rectifies the drawbacks of conventional ELM and other classical learning algorithms. New brain image dataset which is public is used for evaluating and comparing the performance of proposed method. Highly accurate results are produced by this method when compared to other available methods as shown by experimental results.

Volume 11 | 11-Special Issue

Pages: 229-236

DOI: 10.5373/JARDCS/V11SP11/20192952