DOFCM- PSO : A novel hybridized fuzzy clustering technique for segmentation of noisy mammogram images

Fuzzy clustering techniques are commonly applied to manage intricate data patterns using fuzzy partitioning methods. However, their performance deteriorates in presence of noise/outliers and higher-dimensional spaces. To address these limitations, we propose a hybrid algorithm DOFCM-PSO which hybridizes the Density-Oriented Fuzzy C Means clustering technique with metaheuristic algorithm, Particle Swarm Optimization to leverage their respective strengths. The proposed algorithm addresses the challenges of noisy datasets in higher dimensional spaces and provides a promising solution. We evaluate and compare the performance of DOFCM-PSO with existing algorithms on three benchmark datasets (Iris, Wine, and Glass), four real-world digital images, and three breast cancer image datasets. The experimental results demonstrate that DOFCM-PSO consistently outperforms all other algorithms. By effectively integrating DOFCM and PSO, our algorithm achieves improved accuracy, robustness, and stability in various applications. The findings highlight the potential of DOFCM-PSO as a valuable tool for real-world applications.