PSO with Constriction Coefficient and EPFCM Clustering Based Segmentation of Medical Images Using Level Set Method

In image processing one of the pivotal steps is the image segmentation. Image-segmentation is deals with the partitioning of image into meaningful regions based on intensities of the pixel values. It helps to analyze the representation of the image more efficiently and a challenging task in medical images because it is concerned with numerous clinical applications. A new medical image segmentation method based on particle swarm optimization (PSO) with constriction coefficient (CC) and enhanced probabilistic Fuzzy C-Means (EPFCM) is proposed in this paper. The PSOCC-EPFCM clustering algorithm uses random pixel values from medical images as search agents. This is known as pre-processing an image, and it enhances clustering performance by selecting optimal random clusters instead of regular cluster centers. In clustered findings, segments acquired during pre-processing are incoherent and excessively noisy. Post-processing is therefore required to successfully reduce noise, boundary leakages (outliers), and retain object characteristics produced by under and over segmentation. For robustness, the segmentation results must be refined in the post-processing stage using level-set approach. The level set method is a sophisticated numerical methodology for image segmentation and image analysis that use the speed function for curve evolution. For reliable medical image segmentation, the suggested Level Set approach is based on PSOCC-EPFCM clustering. The suggested segmentation method's applicability and efficacy are demonstrated by its use on typical medical images such as MRI and CT scans. The suggested algorithm's performance has effectively offered superior values of Dice Similarity (DS), Jaccard Similarity (JS), and Hausdorff distance (HD) over existing level set approaches.