Improving Diagnosis of Alzheimer's Disease Using Pattern Analysis of Brain Imaging Data

Project Description

Functional Magnetic Resonance Imaging (fMRI) is a powerful tool to measure changes in blood oxygenation level in response to neural activity - when a brain area is more active it consumes more oxygen. Learning from spontaneous fluctuations in the blood oxygenation level-dependent signal measured at resting state functional magnetic resonance imaging have greatly advanced our understanding of how cortical regions interact in large-scale systems. Consistent spatial patterns of some brain networks have been reliably identified in multiple populations [1] [2]. Despite fMRI data provide high spatial and temporal resolution to analyze brain activity, its foundation highly depend relies on the employed statistical analysis and signal modeling methods. The fMRI pattern analysis imposes great challenges in big data analytics and pattern recognition. In this project, we will investigate fMRI data pattern analysis methods, and extend the discriminative pattern analysis to the application of Parkinson’s disease (PD) diagnosis using fMRI data.

In this project, we will work with the brain imaging dataset provided by University of Washington, Integrated Brain Imaging Center. The dataset contains 25 PD patients and 21 healthy subjects. The fMRI data have been prepressed and reduced to 96 time series of brain activation, according to the Harvard-Oxford Brain Atlas with 96 brain cortical regions [3]. The sampling rate of the fMRI data is 2Hz that is one image was produced every 500 milliseconds. The goal of this project is to investigate data mining and pattern recognition methods to improve the pattern analysis of fMRI data in the diagnosis of brain disorder. In particular, the project will focus on multi-voxel pattern analysis (MVPA), which has gained increasing interest in computational neuroscience [4]. MVPA methods have been shown effective to discriminate brain conditions with higher sensitivity than conventional univariate analysis methods. MVPA methods represent a set of possible classifiers or learning machines to decoded fMRI activity patterns and associate learned patterns to different brain conditions. In previous work, we have already constructed a pattern classification framework, which has three major components: feature extraction, feature selection, and classification model construction. The students can study and select one of the areas as their research focus to perform literature review, method investigation, and performance evaluation. The goal of this project is to investigate and implement some new computational methods or algorithms into the current pattern classification framework to improve the diagnostic performance of Alzheimer's Disease using fMRI brain imaging data.

Project Advisor:

Shouyi Wang, Ph.D.

Assistant Professor

Department of Industrial and Manufacturing Systems Engineering

University of Texas at Arlington

500 West First Street, Arlington, TX 76019

Office: Woolf Hall 420H

Tel: 817-272-2921

Fax: 817-272-3406

Email:

References

[1] C. Beckmann, M. DeLuca, J. Devlin, and S. Smith, “Investigations into resting-state connectivity using independent component analysis.” Philos.Trans. R. Soc. Lond. B. Biol. Sci. , vol. 360, p. 1001–1013, 2005.

[2] J. Damoiseaux, S. Rombouts, F. Barkhof, P. Scheltens, C. Stam, S. Smith, and C. Beckmann, “Consistent resting-state networks across healthy subjects.” Proc. Natl. Acad. Sci. USA , vol.103, p. 13848–13853., 2006.

[3] J. Power, A. Cohen, S. Nelson, A. Wig, K. Barnes, J. Church, A. Vogel, T. Laumann, F. Miezin, B. Schlaggar, and S. Petersen, “Functional network organization of the human brain.” Neuron , vol. 72, pp. 665–678, 2011.

[4] MVPA Matlab Toolbox: http://www.brainvoyager.com/bvqx/doc/UsersGuide/MVPA/MultiVoxelPatternAnalysisMVPA.html