An Efficient Segmentation of Biomedical Images using Optimized Level Set Method via Enhanced Possibilistic FCM Clustering

Abstract

The important aspect to deal with medical images is image-segmentation. Image-segmentation is the way to remove the vicinity of attentiveness using different image segments. This segmentation helps to analyze the representation of the image more efficiently. The segmentation of the image in medical image analysis is taken as a challenging task because it is concerned with numerous clinical applications. To have a segmentation of colonoscopy images, cardiac vascular, knee image and MRI (Magnetic Resonance Imaging) brain images', a new approach in this paper is put forward by combining the particle swarm optimization (PSO) and enhanced possibilistic Fuzzy C-Means algorithm. The proposed algorithm proceeds in two stages: In the first stage, optimum pixel values are calculating automatically from the input medical images by using the PSO algorithm. These optimum pixels are acted as a random cluster centres for enhanced Possibilistic Fuzzy C-Means (EP-FCM) clustering algorithm. This process improves the clustering efficiency because of optimum random clusters are choosing instead of normal cluster centres. Segments obtained in the pre-processing are incoherent and highly noisy in clustered results. Therefore, Post-processing is necessary to effectively reduced the noise and boundary leakages (outliers’). In the Second Stage, it is necessary to refine the segmentation results in the pre-processing stage by using the level-set method for robustness. In this paper, tested proposed segmentation algorithm on medical images such as MRI brain and CT Colonoscopy images. The performance of the algorithm is proven to be outstanding with the best accuracy and has dealt with noise effect, boundary leakages effectively.