Robust Fuzzy C-Means in Classifying Breast Tissue Regions

Limited spatial resolution, poor contrast, overlapping intensities, noise and intensity inhomogeneities variation make the assignment of segmentation of medical images is greatly difficult. In recent days, mathematical algorithm supported automatic segmentation system plays an important role in segmentation of medical imaging. This paper presents an effective fuzzy segmentation algorithm for breast magnetic resonance imaging data. This paper obtains a new effective objective function of fuzzy c-means called Kernel induced Fuzzy C-Means based hyper tangent function based on Kernel functions, hyper tangent functions, and Lagrangian multipliers. Initially, this paper tries to derive new objective function by introducing kernel function and consequently it derives new effective equations for calculating memberships and updating prototypes, which are shown to be more robust than FCM. The performance of proposed method has been shown with random data and then the new approach is applied to real medical images. Experimental results on both basic FCM and proposed method have been done for the purpose of comparison of proposed method's result.