Fuzzy C Means is one of the most popular machine learning technique for image segmentation. However, traditional Fuzzy C Means is insensitive to noise as it does not consider spatial information. To solve this issue a wide variety of modified Fuzzy C means techniques, considering spatial information of pixels, are proposed by different researchers. In this paper we propose a hierarchical Fuzzy C Means algorithm considering spatial features of image pixels. Our method aims to overcome the shortcomings of traditional Fuzzy C Means by incorporating spatial feature as well as the issue of misclassification of pixels associated with single level clustering. The proposed method divides the original image pixels into a set of clusters using a spatial fuzzy C means technique in the first level of the hierarchical model. In the second level of the hierarchy, the cluster which contains the candidate mass is further divided into sub clusters using traditional Fuzzy C Means algorithm to yield the final segmentation result. The experimental outputs show improved segmentation result by our proposed method.

Clustering, Fuzzy, Spatial, Segmentation, Hierarchical

[1] “Breast Cancer Statistics”, Worldwide Data, World Cancer Research Fund, USA, 2018
[2] “American College of Radiology (ACR): ACR Breast Imaging Reporting and Data System”, Breast Imaging Atlas, edn. 4, USA, 2003.
[3] A. W. C. Liew, H. Yang, H. F. Law, “Image segmentation based on adaptive cluster prototype estimation”, IEEE Trans. Fuzzy Syst., Vol.13, Issue4, pp.444–453,2005.
[4] H. Zhou, G. Schaefer, C. Shi, “A mean shift based fuzzy c-means algorithm for image segmentation”, In the Proceeding of the 30th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Canada, pp. 3091–3094, 2008.
[5] J. Anitha, J. D. Peter, “Mass segmentation in mammograms using a kernel-based fuzzy level set method”, Int. J. Biomedical Engineering and Technology, Vol. 19, Issue.2, pp. 133-153,2015.
[6] J. C. Dunn, “A fuzzy relative of the ISODATA process and its use in detecting compact well separated clusters”, Journal ofCybernetics, Vol3, Issue. 3, pp.32–57, 1974
[7] J. Bezdek, “Pattern Recognition with Fuzzy Objective FunctionAlgorithms”, Springer, United States, pp. 1981.
[8] X. Y. Wang, J. Bu, “A fast and robust image segmentation usingFCM with spatial information”, Digital Signal Processing, Vol. 20, Issue. 4, pp. 1173–1182, 2010.
[9] W. Cai, S. Chen, D. Zheng, “Fast and Robust Fuzzy C-Means Clustering Algorithms Incorporating Local Information for Image Segmentation”, Pattern Recognition, Vol. 40, Issue. 3, pp. 825-838, 2007.
[10] S. Krinidis, V. Chatzis, “A Robust Fuzzy Local Information C Means Clustering Algorithm”, IEEE Trans Image Process, Vol. 19, Issue. 5, pp. 1328–1337, 2010.
[11] L. Szilagyi, Z. Benyo, S. Szilagyii, and H. Adam, “MR brain image segmentation using an enhanced fuzzy C-means algorithm”, In the Proceedings of 25th Annual International Conference of IEEE EMBS, Mexico, pp. 17–21, 2003.
[12] M. Ahmed, S. Yamany, N. Mohamed, A. Farag, and T. Moriarty,“A Modified Fuzzy C-Means Algorithm for Bias Field Estimation and segmentation of MRI Data”. IEEE Transactions on Medical Imaging, Vol. 21, Issue. 3,pp. 193–199, 2002.
[13] Z. Yuhui, B. Jeon, Q. M. J. Wu, “Image segmentation by generalized hierarchical fuzzy C-means algorithm”, Journal of Intelligent & Fuzzy Systems: Applications in Engineering and Technology, Vol. 28, Issue 2, pp. 961-973, 2015.
[14] J. Suckling et al., “The mammographic image analysis society digital mammogram database”, In the Proceeding of 2nd International Workshop on Digital Mammography, Elsevier Science, England, pp 375–378, 1994.