查看文章

As the field of medical imaging is evolving dependency on automated ways have increased. Many new techniques have been suggested and implemented for the better classification of normal and abnormal brain images. Perfect segmentation of tumor affected portion is one of the challenging tasks, because not only tumor of each patient varies in size, type and location. In addition, effect of noise and non-uniformity in the data during image acquisition through MRI adds on more challenging factor. In this paper we propose a model to classify the tumorous and non-tumorous brain by utilizing the Particle Swarm Optimization (PSO) based segmentation, relevant features and SVM classifier. PSO segments the exact tumor location from the images and extracted thirteen different features with Discrete Wavelet Transform (DWT) based features. These features are trained on SVM classifier with two different kernel …