Title: An Approach for Discretization and Feature Selection Of Continuous Valued Attributes in Medical Images for Classification Learning.
Project Intro: Medical images are a fundamental part of medical diagnosis and treatment. These images are different from typical photographic images primarily because they reveal internal anatomy as opposed to an image of surfaces. They include both projection x-ray images and cross-sectional images, such as those acquired by means of computed tomography (CT) or magnetic resonance imaging (MRI), or one of the other tomography modalities (SPECT, PET, or ultrasound, for example). Medical image processing is a branch of image processing that deals with such images. It is driven both by the peculiar nature of the images and by the medical applications that make them useful. Medical images contain a wealth of hidden information that can be exploited by physicians in making reasoned decisions about a patient. However, extracting this relevant hidden information is a critical initial step to their use. This motivates the use of data mining techniques for efficient knowledge extraction.
Abstract: Many supervised machine learning algorithms require a discrete feature space. In this paper, we review previous work on continuous feature discretization and, identify defining characteristics of the method. We then propose a new supervised approach which combines discretization and feature selection to select the most relevant features which can be used for classification purpose. The classification technique to be used is Associatve Classifiers. The features used are Harlick Texture features extracted from MRI Images. The results show that the proposed method is efficient and well-suited to perform preprocessing of continuous valued attributes.
Keywords: Classifier ,Discretization , Feature Selection, MRI.
You can also Subscribe to PROJECTSWORLDS by Email for more such projects and seminar.
