Classification of functional brain images using a GMM-based multi-variate approach

• Published in: Neuroscience letters Vol. 474; no. 1; pp. 58 - 62
• Main Authors: Segovia, F., Górriz, J.M., Ramírez, J., Salas-González, D., Álvarez, I., López, M., Chaves, R., Padilla, P.
• Format: Journal Article
• Language: English
• Published: Shannon Elsevier Ireland Ltd 19.04.2010; Elsevier
• Subjects: Alzheimer Disease - diagnostic imaging; Alzheimer Disease - physiopathology; Alzheimer's disease; Analysis of Variance; Artificial Intelligence; Biological and medical sciences; Brain - diagnostic imaging; Brain - physiopathology; Degenerative and inherited degenerative diseases of the nervous system. Leukodystrophies. Prion diseases; EM algorithm; Functional brain images; Fundamental and applied biological sciences. Psychology; Gaussian mixture models; Humans; Medical sciences; Models, Statistical; Neurology; Pattern Recognition, Automated; PET; Positron-Emission Tomography; Radiography; Reference Values; SPECT; Tomography, Emission-Computed, Single-Photon; Vertebrates: nervous system and sense organs; Gaussian mixture models; Alzheimer's disease; EM algorithm; Functional brain images; SPECT; PET; Nervous system diseases; Alzheimer disease; Central nervous system; Photon; Cerebral disorder; Encephalon; Central nervous system disease; Degenerative disease; Models; Emission tomography
• ISSN: 0304-3940; 1872-7972; 1872-7972
• DOI: 10.1016/j.neulet.2010.03.010

This paper presents a novel method for automatic selection of regions of interest (ROIs) of functional brain images based on Gaussian mixture models (GMM), which relieves the so-called small size sample problem in the classification of functional brain images for the diagnosis of Alzheimer's disease (AD). In a first step, brain images are preprocessed in order to find an average image including differences between controls and AD patients. Then, ROIs are extracted using a GMM which is adjusted by using the expectation maximization (EM) algorithm. This reduced set of features provides the activation map of each patient and allows us to train statistical classifiers based on support vector machines (SVMs). The leave-one-out cross-validation technique is used to validate the results obtained by the supervised learning-based computer aided diagnosis (CAD) system over databases of SPECT and PET images yielding an accuracy rate up to 96.67%.