Segmentation of Tumors in MRI Brain Images using Modified PSO and ADF based SVM

Abstract

poor illuminations, intensity inhomogeneity and the noise are the biggest challenging task in the medical image segmentation. The segmentation of biomedical images is considered as one of the stimulating tasks in many clinical applications. In the last decades, researchers proposed many methods and systems where it can do the automatic segmentation of an medical images without any interference. In this paper presented a new hybrid algorithm for accurately identifying and segmenting tumors in medical images by effective utilization of Modified Particle Swarm Optimization (MPSO) and anisotropic diffusion filter based Support Vector Machine (ADF-SVM). The proposed method has been diversified into two stages. In the initial stage, a hybrid approach is considered by combining modified Particle Swarm Optimization (MPSO) and Anisotropic Diffusion Filter (ADF) which is called MPSO-ADF is employed on MR images in order to eradicate the noise and smoothing the input medical images by preserving edge information. In the next step, Support Vector Machine Classifier (SVM) is utilized in lieu of the perfect identification and segmenting tumors from the refined images. The proposed method has been demonstrated on various medical images and analyzed the segmentation results in terms of dice coefficient, Jacquard coefficient over conventional filtering techniques. The proposed method has given good segmentation results over existing methods in terms of dice and jaccard coefficients respectively.