Abstract
In biomedical domain, magnetic resonance imaging (MRI) segmentation is
highly essential for the treatment or prevention of disease. The demand for fast
processing and high accurate results is necessary for medical diagnosis. This can be
solved by using computational intelligence (CoIn) for data processing. The CoIn can be
achieved by using well-known techniques such as fuzzy logic, genetic algorithm,
evolutionary algorithms and neural networks. The computational complexity of a
medical image segmentation depends on the characteristics of data as well as suitable
algorithms. The selection of CoIn methods is very important for better segmentation of
a medical image because each algorithm outperforms a different medical image data
set. The hybrid CoIn (H-CoIn) is one of the solutions to overcome the problem of
individual algorithms in medical image segmentation. The H-CoIn is a combination of
two or more intelligence algorithms (like fuzzy logic, evolutionary algorithms and
neural networks). The drawbacks of individual intelligence algorithms can be
overcome by using H-CoIn. In a medical image segmentation process, two or more
variables or objectives need to be optimized for H-CoIn. This problem can be solved by
using multi-objective optimization techniques, where simultaneously minimization or
maximization can be performed. In this chapter, the various CoIn algorithms'
performance has been discussed in detail for medical image segmentation and
compared with state-of-the-art techniques. The H-Coin algorithm has been
implemented in a large medical dataset and attained an accuracy of 98.89%. Further,
the H-Coin algorithm is reliable and suitable to overcome the inter-observer and intraobserver variability.