Chapter 6 - Computational intelligence techniques for cancer diagnosis

Abstract

Computational intelligence (CI) is a group of techniques for building machine intelligence. These techniques follow a pragmatic approach to learning and decision making rather than a hard approach as in expert systems or rule-based systems. These techniques use soft computing, fuzzy logic, and evolutionary methods to enable a machine to learn and make a decision under different situations. This is different from artificial intelligence which is based on crisp logic rules. In this chapter, we discuss the applications of important techniques under CI that have been applied to the computational diagnosis of cancers. These important techniques include fuzzy logic, artificial neural networks, evolutionary computation based on principles of natural selection, learning theory which is the study of learning mechanism of natural organisms, probabilistic or random methods that inherently account for uncertainty in input or events. Fuzzy logic is applied in the identification of tumors in medical imaging reports where approximate reasoning is helpful. Neural networks like convolutional neural networks have been demonstrated to accurately identify and classify the various type of tumors. An evolutionary computation or natural computation methods apply the principle of natural selection to solve a multiobjective optimization problem. These methods, including swarm intelligence, etc., have wide applications in the areas of genomic data analysis. CI heavily uses the principles of learning theory to understand the cognitive processes of various natural organisms. Learning theory involves the study of how the processes of cognition, emotion regulation, accounting for environmental influences, and experience aid in gathering, improving, or changing knowledge. It also examines how these are processed and used for the prediction of future events based on experience. The application of learning theory to the prediction of the incidence of a disease is an interesting area of research. Probabilistic methods are characterized by randomness. Many probabilistic methods like Bayesian networks are used in the diagnosis of cancers. The mathematical basis is mainly machine learning methods like regression, support vector machines, decision trees, etc. In this chapter, we have analyzed commonly used CI techniques as applied to research into the classification and diagnosis of cancerous tumors and compiled a list of the most promising techniques that have improved the accuracy of diagnosis or have led to better outcomes for cancer patients.