Constructing Brain Connectivity Graph by Modified Sparse Representation
In the field of neuroimaging, fMRI is an important tool for brain connectivity analysis. However, the architecture of functional connectivity within the human brain connectome cannot be exactly interpreted at the voxel level by using the traditional correlation analysis. To address this problem, we propose a modified sparse representation (MSR) method to construct the connectivity graph in an automatical and efficient way. The MSR approach uses the sparse representation instead of the correlation coefficient to relate brain regions or voxels. Degree centrality (DC), closeness centrality (CC), betweenness centrality (BC), and eigenvector centrality (EC) are employed to extract the features of fMRI connective patterns. With the extracted features, we then experimentally compare affirmative and negative sentences processing on the Star/Plus database, which shows significant difference via MSR method. Compared with the traditional correlation method, MSR shows higher significance between the two cognitive processing tasks.
Keywordsmodified sparse representation degree centrality closeness centrality betweenness centrality eigenvector centrality
Unable to display preview. Download preview PDF.
- 2.Horwitz, B., Horovitz, S.G.: Introduction to Research Topic–Brain Connectivity Analysis: Investigating Brain Disorders. Part 1: The Review Articles. Frontiers in Systems Neuroscience 6, 1–2 (2012)Google Scholar
- 5.Zhang, J., Cheng, W., Wang, Z.G.: Pattern Classification of Large-scale Functional Brain Networks: Identification of Informative Neuroimaging Markers for Epilepsy. PLoS One 7, e36733 (2012)Google Scholar
- 8.Donoho, D.L., Tsaig, Y.: Fast Solution of l1-norm Minimization Problems When the Solution May Be Sparse. Department of Statistics, Stanford University (2006)Google Scholar
- 9.Lohmann, G., Margulies, D.S., Horstmann, A.: Eigenvector Centrality Mapping for Analyzing Connectivity Patterns in fMRI Data of the Human Brain. PLoS One 5, e10232 (2010)Google Scholar
- 10.Druckmann, S., Chklovskii, D.B.: Over-complete Representations on Recurrent Neural Networks Can Support Persistent Percepts. Advances in Neural Information Processing Systems 23, 541–549 (2010)Google Scholar
- 11.Richiardi, J., Van De Ville, D., Eryilmaz, H.: Low-dimensional Embedding of Functional Connectivity Graphs for Brain State Decoding. In: 2010 First Workshop on. 2010 Brain Decoding: Pattern Recognition Challenges in Neuroimaging (WBD). IEEE (2010)Google Scholar
- 16.Napoletani, D., Sauer, T.D.: Reconstructing the Topology of Sparsely Connected Dynamical Networks. Physical Review-Section E-Statistical Nonlinear and Soft Matter Physics 77, 26103 (2008)Google Scholar