Intelligent System for Brain Disease Diagnosis Using Rotation Invariant Features and Fuzzy Neural Network

The characteristic features of the magnetic resonant image (MRI) for Alzheimer's patient's brain image and normal image can be distinguished in terms of dimensional features with the help of wavelet decomposition. From the literature review, it is observed that when datasets used are a combination of the MR images having a very mild cognitive impairment and mild cognitive impairment, the performance of the classifier reduces. Because the features of this kind of MR image are difficult to distinguish from normal brain images. To solve this problem, the lossless feature extraction method along with the feature reduction method having a selection approach is suggested as a solution here. In this paper, the 12 directional, rotation invariant two-dimensional discrete-time continuous wavelet transform (R-DTCWT) and a genetic algorithm (GA) are used for feature selection and feature vector size reduction. The fuzzy neural network (FNN) which is suitable for pattern recognition is used here. The FNN with and without feature reduction is evaluated for identification of combinational dataset, shows satisfactory performance over an artificial neural network (ANN), probabilistic neural network (PNN) classifiers. This method is compared with other state of algorithm to prove the enhanced performance