Image segmentation has been an intriguing area for research and development of efficient algorithms, playing a key role in the interpretation of high-caliber images and image analysis. Segmentation of images plays an imperative role in medical diagnosis. Such segmentation requires a robust algorithm of segmentation against noise. The legendary orthodox fuzzy c-means algorithm is exploited with domain to be grouped into the segmentation of medical images. FCM is highly sensitive to noise due to the practice of only intensity values for clustering. Therefore, this work aims to apply the "kernel method", instituted in the conventional fuzzy clustering algorithm (FCM) to exchange the Euclidean metric with a novel metric induced by the kernel in the data space. Images can be segmented by pixel classification by grouping all the characteristics of interest. In non-supervised clustering algorithm methods that use the kernel method, a nonlinear mapping is initially operated in order to map the data to a much larger space characteristic, and then clustering is performed. The cluster integer in the multidimensional entity space thus represents the number of classes in the image. As the image is sorted into clustered classes, the segmented regions are obtained by examining the neighbourhood pixels for the same class tag. Since clustering produces disjoint regions with holes or regions with a single pixel, a post-processing algorithm such as growing region, pixel connectivity, or a rule-based algorithm is applied to obtain the final segmented regions.