Nonlinear classification methods for MRI

Much attention is currently focused on newest brain examination techniques, brain MRI. However, segmentation of magnetic rezonance images is a challenging problem due to their complexity. Enhanced brain MRIs acquired by nonlinear classification methods, especially by neural networks have been shown to be very sensitive in the detection of brain tumors, but are also time-consuming and cause waste of medical resources.This thesis therefore, examines the usage of neural networks detection technology on MR image classification. A series of experiments was conducted on classification of images by K-nearest neighbor Classifier, Parzen Classifier, Linear perceptron Classifier, Support vector Classifier, Construct binary decision tree classifier and the results compared for performance evaluation. After comparison with the algorithms the experimental results showed success of each algorithms related to changing parameters.Keywords: Neural Networks, Clustering, Brain, Image Processing