A robust segmentation approach for noisy medical images using fuzzy clustering with spatial probability

Image segmentation plays a major role in medical imaging applications. During last decades, developing robust and efficient algorithms for medical image segmentation has been a demanding area of growing research interest. The renowned unsupervised clustering method, Fuzzy C-Means (FCM) algorithm is extensively used in medical image segmentation. Despite its pervasive use, conventional FCM is highly sensitive to noise because it segments images on the basis of intensity values. In this paper, for the segmentation of noisy medical images, an effective approach is presented. The proposed approach utilizes histogram based Fuzzy C-Means clustering algorithm for the segmentation of medical images. To improve the robustness against noise, the spatial probability of the neighboring pixels is integrated in the objective function of FCM. The noisy medical images are denoised, with the help of an effective denoising algorithm, prior to segmentation, to increase further the approach’s robustness. A comparative analysis is done between the conventional FCM and the proposed approach. The results obtained from the experimentation show that the proposed approach attains reliable segmentation accuracy despite of noise levels. From the experimental results, it is also clear that the proposed approach is more efficient and robust against noise when compared to that of the FCM.