An Empirical Performance Analysis of Brain Image Classification Models Using Variants of Neural Networks

This paper presents the empirical comparison of different neural network models for the classification of brain Magnetic Resonance Images (MRIs). This work comprises of four stages i.e. dataset collection, feature extraction, feature reduction and classification. The two brain MRI datasets i.e. the Glioma and Alzheimer datasets are considered for this work. Discrete wavelet transformation (DWT) technique is used for the extraction of features from brain MRIs. Principal Component Analysis (PCA) technique is used to for feature reduction to get relevant features. For the classification task, two variants of neural networks i.e. Backpropagation Neural Network (BPNN) and Extreme Learning Machine (ELM) are used and the classification performances are compared using different performance measures. The simulation study exhibits that DWT+PCA+ELM model outperformed the other models for the classification of normal and diseased brain for the two datasets.