Contourlet Transform and PNN Based Brain Tumor Classification

Abstract

In this paper a new method is proposed for Brain Tumor classification using a Probabilistic neural network with contourlet transformation. The conventional methods like Computerized Tomography (CT) and Magnetic Resonance Images (MRI) are based on human inspection for tumor classification and detection. These methods are inefficient and are also non reproducible for large amount of data. Also, those methods include noise due to operator performance which creates serious inaccuracies in tumor classification. Neural Networks shows good results in medical diagnosis which is combined here with contourlet transform. Decision making is based on two steps (i) Image reduction and Feature extraction using contourlet transform (ii) Classification using probabilistic Neural Network(PNN). Performance evaluation on various brain tumor images shows fast and better recognition rate and also low computational requirements, when compared to previous classifiers.