UBC Theses and Dissertations
Joint source based brain imaging analysis for classification of individuals Ramezani, Mahdi
Diagnosis and clinical management of neurological disorders that affect brain structure, function and networks would benefit substantially from the development of techniques that combine multi-modal and/or multi-task information. Here, we propose a joint Source Based Analysis (jSBA) framework to identify common information across structural and functional contrasts in data from MRI and fMRI experiments, for classification of individuals with neurological and psychiatric disorders. The framework consists of three components: 1) individual's feature generation, 2) joint group analysis, and 3) classification of individuals based on the group's generated features. In the proposed framework, information from brain neuroimaging datasets is reduced to a feature that is a lower-dimensional representation of a selected brain structure or task-related activation pattern. For each individual, features are used within a joint analysis method to generate basis brain activation sources and their corresponding modulation profiles. Modulation profiles are used to classify individuals into different categories. We perform two experiments to demonstrate the potential of the proposed framework to classify groups of subjects based on structural and functional brain data. In the fMRI analysis, functional contrast images derived from a study of auditory and speech perception of 16 young and 16 older adults are used for classification of individuals. First, we investigate the effect of using multi-task fMRI data to improve the classification accuracy. Then, we propose a novel joint Sparse Representation Analysis (jSRA) to identify common information across different functional contrasts in data. We further assess the reliability of jSRA, and visualize the brain patterns obtained from such analysis. In the sMRI analysis, features representing position, orientation and size (i.e. pose), shape, and local tissue composition of brain are used to classify 19 depressed and 26 healthy individuals. First, we incorporate pose and shape measures of morphology, which are not usually analyzed in neuromorphometric studies, to measure structural changes. Then, we combine brain tissue composition and morphometry using the proposed jSBA framework. In a cross-validation leave-one-out experiment, we show that we can classify the subjects with an accuracy of 67% solely based on the information gathered from the joint analysis of features obtained from multiple brain structures.
Item Citations and Data
Attribution-NonCommercial-NoDerivs 2.5 Canada