A Statistical Method to Distinguish Functional Brain Networks

Autor(es)

Fujita, Andre ; Vidal, Maciel C. ; Takahashi, Daniel Y.

• Tipo de documento: Artigo Científico
• Fonte: FRONTIERS IN NEUROSCIENCE; v. 11, FEB 14 2017.
• Citações Web of Science: 8

Resumo

One major problem in neuroscience is the comparison of functional brain networks of different populations, e.g., distinguishing the networks of controls and patients. Traditional algorithms are based on search for isomorphismbetween networks, assuming that they are deterministic. However, biological networks present randomness that cannot be well modeled by those algorithms. For instance, functional brain networks of distinct subjects of the same population can be different due to individual characteristics. Moreover, networks of subjects from different populations can be generated through the same stochastic process. Thus, a better hypothesis is that networks are generated by random processes. In this case, subjects from the same group are samples from the same random process, whereas subjects from different groups are generated by distinct processes. Using this idea, we developed a statistical test called ANOGVA to test whether two or more populations of graphs are generated by the same random graph model. Our simulations' results demonstrate that we can precisely control the rate of false positives and that the test is powerful to discriminate random graphs generated by different models and parameters. The method also showed to be robust for unbalanced data. As an example, we applied ANOGVA to an fMRI dataset composed of controls and patients diagnosed with autism or Asperger. ANOGVA identified the cerebellar functional sub-network as statistically different between controls and autism (p < 0.001). (AU)

• Processo FAPESP: 13/03447-6 - Estruturas combinatórias, otimização e algoritmos em Teoria da Computação
• Beneficiário: Carlos Eduardo Ferreira
• Linha de fomento: Auxílio à Pesquisa - Temático
• Processo FAPESP: 13/07375-0 - CeMEAI - Centro de Ciências Matemáticas Aplicadas à Indústria
• Beneficiário: José Alberto Cuminato
• Linha de fomento: Auxílio à Pesquisa - Centros de Pesquisa, Inovação e Difusão - CEPIDs
• Processo FAPESP: 15/01587-0 - Armazenagem, modelagem e análise de sistemas dinâmicos para aplicações em e-Science
• Beneficiário: João Eduardo Ferreira
• Linha de fomento: Auxílio à Pesquisa - Programa eScience e Data Science - Temático