Brain Tumor Classification using Discrete Curvelet Transform and Probabilstic Neural Network

In this paper, an automatic support system for brain tumor stage classification using learning machine and for detecting brain tumor during early stages using fuzzy clustering methods is proposed. The fuzzy clustering method is a segmentation technique presented to segment the Magnetic Resonance images for detecting the Brain Tumor during early stages and for examining anatomical structures. Fast discrete curvelet transform is proposed to examine the texture of the image. The learning machine approach is used to classify the stage of Brain Tumor that is benign, malignant or normal. The automated brain tumor classification can be implemented in two stages: feature extraction using Gray-Scale Co-occurrence matrix (GLCM) and the classification using PNN-RBF network. The simulated results show that proposed classifier and segmentation algorithm provides better accuracy than previous methods.