ABSTRACT
In medical research and development, the computational domain provides support and calculative power to perform unique and innovative tasks, among which image categorization is a huge and an important issue. In the world of medical science, every day doctors encounter with various types of image-related problems. Pattern classification is a solution to categorize images. In the human body, the brain plays a vital role, and the sudden growth of anomalous tissues in the brain can cause brain cancer. When unusual cells and tissues grow relatively slowly, the resulting tumors are classed as benign, which indicates they are non-cancerous, whereas in malignant tumors, the cells grow very rapidly and cause serious harm to the brain, often leading to death. There are many methodologies for producing medical images of the brain. Among them, magnetic resonance imaging is a very powerful technique for producing digital images of internal slices of the brain. This information is very helpful for medical diagnosis and research. To identify and classify a tumor in the human body, the first process is image acquisition, then image segmentation techniques are used to explore the anomalous tissues. The method proposed in this chapter is motivated by the need for improved and quicker medical image analysis for brain tumors. In this image segmentation approach, Otsu’s method is used along with principal component analysis to select the appropriate features for extraction. The gray level co-occurrence matrix method is used to calculate some valuable texture parameters, and finally, machine learning techniques and kernel functions are used to achieve high accuracy with the help of support vector machines.
