Brain tumour detection in MRI images using PNN and GRNN

Brain tumor detection is the most significant method to describe the early tumor. Enlarging the tumor is being a huge challenge due to the complex characteristics of the MRI Images which gives high intensive, divergent and uncertain boundaries. To address this problem, tumor segmentation method for MRI images which separates tumorous cells from healthy tissues has been carried out by the use of two types of clustering methods. In the proposed method input Image is pre-processed, followed by which the segmentation is done using K Means clustering method and Fuzzy C Means clustering method. While comparing these two techniques, it is seen that Fuzzy C Means clustering produces better segmentation. Further the features like magnitude, direction and the area are extracted from the tumorous part of Fuzzy C Means segmented Image. Based on the features extracted, the MRI image is classified as tumorous or non-tumorous. Classification is done by using the supervised neural network called the Radial Basis Function (RBF), Generalised Regression neural network (GRNN), Probabilistic neural network (PNN). In the proposed method the classification is done by using the Fuzzy probabilistic neural network classifier (FPNNC) has been used to classify the MRI Image as normal or abnormal and the classifiers are compared in terms of accuracy, specificity and sensitivity. This proves that the proposed method produces better accuracy compared to the existing classifiers.