Abstract: In this paper Brain cancer detection & classification system has been designed & developed for distinguishing different types of brain MRI into three classes such as Benign, Malignant and Normal. The image processing techniques such as image acquisition, image segmentation, morphological operations & feature Extraction have been developed for detection of brain tumor. In this project we have obtained the features related to the Discrete Cosine Transform as well as Discrete Wavelet Transform. For segmentation of tumor region K-means clustering is used. The extraction of texture features in the detected tumor has been achieved by using Gray level co-occurrence matrix (GLCM). Probabilistic Neural Network is employed to implement an automated brain tumor classification. The performance of the PNN classifier is evaluated in terms of training performance and classification accuracies. Tumor area is calculated. Finally there is comparison between accuracy analysis for Discrete Cosine Transform (DCT) and Discrete Wavelet Transform (DWT) over 90 MRI of brain. Simulation is performed by MATLAB software.

Keywords: Discrete Cosine Transform(DCT), Discrete Wavelet Transform (DWT), Gray level co-occurrence matrix(GLCM), K-means Clustering, MATLAB, Magnetic Resonance Images(MRI), Probabilistic Neural Network.