Clustering of Medical Images - Medical Image Processing: Advanced Fuzzy Set Theoretic Techniques - page 160

The membership matrix will follow Equation 7.14 using hypertangent ker-
nel, but hesitation degree is incorporated into each membership degree.
Kernel-based IFCM clustering is robust on noise. It works well in a noisy
environment. Examples are shown on tumour images using all the three
kernels.

E x a mple 7. 5

Two examples are shown on CT scan images of the brain to show the
efficacy of the algorithm. Figures 7.7 and 7.8 show the tumour/clot in the
brain.

It is observed that simple intuitionistic fuzzy clustering does not per-
form better in a noisy environment but works well in a noisy environ-
ment. Radial basis kernel does not work well. Hypertangent kernel works
a little bit better than Gaussian kernel where the tumour/clot is clearly
detected.

(a) (b) (c)

(d) (e)

FIGURE 7.7
(a) Brain clot image, (b) IFCM, (c) IFCM with Gaussian kernel, (d) IFCM with hypertangent ker-
nel and (e) IFCM with radial basis kernel. (Modified from Chaira, T. and Panwar, A., J. Comput.
Intell. Syst. , 7(2), 1-11, 2013.)