Duck Cluster Optimization Algorithm with K-Means Clustering for Mammogram Image Segmentation

Abstract

Objective: The uncertainty of uncontrollable cell growth of breast in women seriously affected and protected from the breast cancer. Breast cancer is the serious cancer in women it required to determine in the earlier stage before the deadline condition. The ultimate innovation of this paper is segmenting breast tumor by using new meta-heuristic algorithm called Duck Cluster Optimization. Methods: K-means Clustering calculation utilizing a meta-heuristic Duck Cluster Optimization (DCO) calculation has been proposed for taking care of the picture division issue. DCO calculation is utilized to amplify the segmentation goal capacities. Particle Swarm Optimization (PSO) and Ant Colony Optimization (ACO) of swarm intelligence algorithms are efficiently used for validating the performance of DTO. Test images retrieved successfully and processed usefully from Mammogram Image Analysis Society (MIAS) database. Findings: The results of metrics values Mean Squared Error (MSE), Peak Signal to Noise Ratio (PSNR), Variation of Information (VoI), and Random Index (RI) compared with ACO, PSO and DCO. PSNR values proved that the proposed DCO leads the optimum threshold values comparing with PSO and ACO algorithms. Thus comparison between existing methods and the DCO having the highest robustness value rather than PSO and ACO.