%0 Journal Article
%T Machine Learning Approaches: From Theory to Application in Schizophrenia
%A Elisa Veronese
%A Umberto Castellani
%A Denis Peruzzo
%A Marcella Bellani
%A Paolo Brambilla
%J Computational and Mathematical Methods in Medicine
%D 2013
%I Hindawi Publishing Corporation
%R 10.1155/2013/867924
%X In recent years, machine learning approaches have been successfully applied for analysis of neuroimaging data, to help in the context of disease diagnosis. We provide, in this paper, an overview of recent support vector machine-based methods developed and applied in psychiatric neuroimaging for the investigation of schizophrenia. In particular, we focus on the algorithms implemented by our group, which have been applied to classify subjects affected by schizophrenia and healthy controls, comparing them in terms of accuracy results with other recently published studies. First we give a description of the basic terminology used in pattern recognition and machine learning. Then we separately summarize and explain each study, highlighting the main features that characterize each method. Finally, as an outcome of the comparison of the results obtained applying the described different techniques, conclusions are drawn in order to understand how much automatic classification approaches can be considered a useful tool in understanding the biological underpinnings of schizophrenia. We then conclude by discussing the main implications achievable by the application of these methods into clinical practice. 1. Introduction Investigating the neurobiological bases of psychiatric disorders requires a large sample studied in a longitudinal perspective from early stages of the diseases. In this context, magnetic resonance imaging (MRI) is the gold-standard technique to explore the anatomical and functional underpinnings of such illnesses [1¨C3]. In order to accurately analyze such large amount of imaging data, automated methods are becoming essential [4]. As outlined by Lao and colleagues [5], to develop an accurate detector of pathology from a set of images, two issues need to be addressed. First, an image analysis methodology is needed in order to extract the most relevant information from the images. Second, a pattern classification method has to be designed to process the extracted information, in order to determine the likelihood of the disease. Feature extraction is aimed at characterizing an object in terms of properties, or features, such as dimensions, shape, color, and texture. Chosen features are those that, when belonging to objects of the same category, or class, are very similar; on the contrary, they should be very different from objects in different categories. The set of features extracted from an object can be considered as a signature which describes the object itself. Features are usually organized in the so-called feature vector, a vector of arbitrary
%U http://www.hindawi.com/journals/cmmm/2013/867924/