Implementation of PNN Based on Extraction of DCT features for Brain Tumor Classification

Abstract: Accurate detection, segmentation and classification of Brain tumor is a challenging task in medical field. This paper proposes a method for brain tumor classification using Probabilistic Neural Network based on textural feature extraction of Magnetic Resonance Images (MRI). The proposed scheme consist of several stages including image acquisition, segmentation of tumor part, feature extraction and classification. This project include clustering algorithm for segmentation of MRI to detect brain tumor. Discrete cosine transform (DCT) is used for the dimensionality reduction and feature extraction. PNN will be useful to classify the stages of brain tumor that is Normal, Benign or Malignant. This can be performed in two stages : 1) Gray level co-occurrence matrix and 2) Classification using PNN based function. Detection and extraction of tumor from MRI scan images of the brain is done by using MATLAB software. Evaluation was performed on image database of 90 MRI of brain.

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