Medical image classification using artificial neural network and spider monkey optimization: An empirical analysis

In this work, a novel hybrid image classification strategy based on Artificial Neural Networks (ANNs) and Spider Monkey Optimization (SMO) is been proposed to classify Magnetic Resonance (MR) brain image as either normal or diseased. Discrete Wavelet Transformation is first applied on MR images for feature extraction using Bi-orthogonal wavelet functions before classification task. SMO is a relatively new swarm based meta-heuristic optimization technique that ensures best optimal solution by keeping good balance between exploration and exploitation. In this proposed classification technique, SMO is used to optimize the performance of ANN by determining the hidden node parameters. The performance of the proposed classifier has been compared with other hybridized classifiers using ANN, Genetic Algorithm, Differential Evolution and Particle Swarm Optimization. All experimental validations are done on three different brain disease datasets, viz., Alzheimer, Glioma and Multiple Sclerosis and the results showed that the proposed ANN-SMO model out performs other hybrid methods and achieved accuracy of 98% in case of Alzheimer dataset and accuracy of 99% in both Glioma and Multiple Sclerosis datasets.